Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "deepseek-ai/DeepSeek-V4.1-Flash" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp8; weights repository https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.30 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $1.20 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.01 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists deepseek-ai/DeepSeek-V4.1-Flash with recommended replacement deepseek-ai/DeepSeek-V4.1-Flash-1017. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed deepseek-ai/DeepSeek-V4.1-Flash from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-deepseek-v4-1-flash:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-1-flash: pricing · nebius-deepseek-v4-1-flash: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists deepseek-ai/DeepSeek-V4.1-Flash with recommended replacement deepseek-ai/DeepSeek-V4.1-Flash-1017. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-deepseek-v4-1-flash:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed deepseek-ai/DeepSeek-V4.1-Flash as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists deepseek-ai/DeepSeek-V4.1-Flash again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "deepseek-ai/DeepSeek-V4.1-Flash-1017" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/nvidia/DeepSeek-V4.1-Flash-1017-NVFP4. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.30 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $1.20 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.006 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 159.4 (90% range 157.0–162.0) · range spans bands · Epoch AI score for DeepSeek V4.1 Flash 1017, model dated Oct 17, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-1-flash-1017: pricing · nebius-deepseek-v4-1-flash-1017: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "deepseek-ai/DeepSeek-V4-Flash-0731" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp8; weights repository https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.14 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.28 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.01 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-flash-0731: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "deepseek-ai/DeepSeek-V4-Flash-0731", raw SHA-256 279526ca433255dc4e19562886346dcbc8c8d34f8bdfaf577b1713ad28135b17: "DeepSeek V4 Flash 0731 is a 1M-context reasoning model designed for coding and agentic workloads."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B.
Specifications only: no pricing, availability, capability or release-date assertion.
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists deepseek-ai/DeepSeek-V4-Flash-0731 with recommended replacement deepseek-ai/DeepSeek-V4.1-Flash-1017. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed deepseek-ai/DeepSeek-V4-Flash-0731 from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-deepseek-v4-flash-0731:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-flash-0731: pricing · nebius-deepseek-v4-flash-0731: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists deepseek-ai/DeepSeek-V4-Flash-0731 with recommended replacement deepseek-ai/DeepSeek-V4.1-Flash-1017. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-deepseek-v4-flash-0731:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed deepseek-ai/DeepSeek-V4-Flash-0731 as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists deepseek-ai/DeepSeek-V4-Flash-0731 again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Retired Nov 11, 2026 · price last observed Nov 10, 2026 (retirement notice ↗)
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "deepseek-ai/DeepSeek-V4-Pro" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region uk-south1; quantization fp8; weights repository https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.75 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.50 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.14 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: not verified · Retired: Nov 11, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists deepseek-ai/DeepSeek-V4-Pro with recommended replacement deepseek-ai/DeepSeek-V4-Pro-0813. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": true. The flag is undocumented (#962) and is not the basis of this record; the dated notice is.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the reviewAfter statement above, which describes the 2026-11-06 review: the notice, re-read at 2026-11-11T02:48:39Z (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1), states the same text and table. Nebius's models_info API still listed deepseek-ai/DeepSeek-V4-Pro as active on 2026-11-10 (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157), after the notice's removal date, with the flavor field "deprecated": true. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date. First reviewed 2026-11-06T16:33:57Z (#996).
Re-reviewed 2026-11-11 (#1098), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 and (#999) reviews: Nebius removed deepseek-ai/DeepSeek-V4-Pro from the models_info API between the 2026-11-10 and 2026-11-11 collections, after the notice's date; the retirement is recorded by retirement:nebius-deepseek-v4-pro:2026-11-11, so this record has no further review date.
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-pro: pricing · nebius-deepseek-v4-pro: availability · 44 more claims about 22 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists deepseek-ai/DeepSeek-V4-Pro with recommended replacement deepseek-ai/DeepSeek-V4-Pro-0813. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-deepseek-v4-pro:2026-11-06 quotes.
Observed removal, after the notice's date, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) still listed deepseek-ai/DeepSeek-V4-Pro as active, with the undocumented flavor field "deprecated": true; the 2026-11-11 collection (retrieved 2026-11-11T04:29:52.407Z, raw SHA-256 1bf4bbf98d85308a0fc77d33f22145baa80d824af824de1ef92f45275de87e99, 32,764 bytes, 23 entries) no longer lists it. The two entries that left are exactly DeepSeek-V4-Pro and GLM-5.1; none was added. A same-day re-collect (Actions run 43041657659) (retrieved 2026-11-11T14:46:41.317Z, raw SHA-256 c1a62624019cc8840784d5f1336b14d6b03b97052d14101f211be100e9cb203a) still does not list it.
retiredAt is 2026-11-11, the day of the first collection that no longer lists the model. Nebius removed it after the notice's November 9 date, between 2026-11-10T04:29:59.019Z (still listed as active) and 2026-11-11T04:29:52.407Z (no longer listed); within that window the time is unknown. It is stored as 2026-11-11T00:00:00Z and shown as the date only, next to the 2026-11-10 last observed price.
The rates kept are the last observed Serverless list prices (the 2026-11-10 record). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists deepseek-ai/DeepSeek-V4-Pro again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #1098.
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "deepseek-ai/DeepSeek-V4-Pro-0813" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region uk-south1; quantization fp8; weights repository https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.32 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.96 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.04 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 155.9 (90% range 154.1–158.0) · Epoch AI score for DeepSeek V4 Pro 0813, model dated Aug 13, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-deepseek-v4-pro-0813: pricing · nebius-deepseek-v4-pro-0813: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "google/gemma-3-27b-it" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/google/gemma-3-27b-it. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.10 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.30 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.04 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 127.5 (90% range 124.3–130.4) · Epoch AI score for Gemma 3 27B, model dated Mar 12, 2025
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-gemma-3-27b-it: pricing · nebius-gemma-3-27b-it: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Retired Nov 11, 2026 · price last observed Nov 10, 2026 (retirement notice ↗)
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "zai-org/GLM-5.1" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/zai-org/GLM-5.1. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.40 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.40 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.26 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: not verified · Retired: Nov 11, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists zai-org/GLM-5.1 with recommended replacement zai-org/GLM-5.3. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": true. The flag is undocumented (#962) and is not the basis of this record; the dated notice is.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the reviewAfter statement above, which describes the 2026-11-06 review: the notice, re-read at 2026-11-11T02:48:39Z (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1), states the same text and table. Nebius's models_info API still listed zai-org/GLM-5.1 as active on 2026-11-10 (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157), after the notice's removal date, with the flavor field "deprecated": true. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date. First reviewed 2026-11-06T16:33:57Z (#996).
Re-reviewed 2026-11-11 (#1098), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 and (#999) reviews: Nebius removed zai-org/GLM-5.1 from the models_info API between the 2026-11-10 and 2026-11-11 collections, after the notice's date; the retirement is recorded by retirement:nebius-glm-5-1:2026-11-11, so this record has no further review date.
Raw API evidence ↗ · provider API · Claims: nebius-glm-5-1: pricing · nebius-glm-5-1: availability · 44 more claims about 22 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists zai-org/GLM-5.1 with recommended replacement zai-org/GLM-5.3. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-glm-5-1:2026-11-06 quotes.
Observed removal, after the notice's date, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) still listed zai-org/GLM-5.1 as active, with the undocumented flavor field "deprecated": true; the 2026-11-11 collection (retrieved 2026-11-11T04:29:52.407Z, raw SHA-256 1bf4bbf98d85308a0fc77d33f22145baa80d824af824de1ef92f45275de87e99, 32,764 bytes, 23 entries) no longer lists it. The two entries that left are exactly DeepSeek-V4-Pro and GLM-5.1; none was added. A same-day re-collect (Actions run 43041657659) (retrieved 2026-11-11T14:46:41.317Z, raw SHA-256 c1a62624019cc8840784d5f1336b14d6b03b97052d14101f211be100e9cb203a) still does not list it.
retiredAt is 2026-11-11, the day of the first collection that no longer lists the model. Nebius removed it after the notice's November 9 date, between 2026-11-10T04:29:59.019Z (still listed as active) and 2026-11-11T04:29:52.407Z (no longer listed); within that window the time is unknown. It is stored as 2026-11-11T00:00:00Z and shown as the date only, next to the 2026-11-10 last observed price.
The rates kept are the last observed Serverless list prices (the 2026-11-10 record). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists zai-org/GLM-5.1 again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #1098.
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "zai-org/GLM-5.2" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region uk-south1; quantization fp4; weights repository https://huggingface.co/zai-org/GLM-5.2. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.40 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.40 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.14 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 153.0 (90% range 151.3–154.8) · Epoch AI score for GLM-5.2, model dated Jun 16, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-glm-5-2: pricing · nebius-glm-5-2: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Model ID: zai-org/GLM-5.3 · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "zai-org/GLM-5.3" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/incoai/GLM-5.3-NVFP4. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.40 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.40 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.26 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 157.1 (90% range 154.9–159.5) · Epoch AI score for GLM-5.3, model dated Aug 14, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-glm-5-3: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "zai-org/GLM-5.3", raw SHA-256 279526ca433255dc4e19562886346dcbc8c8d34f8bdfaf577b1713ad28135b17: "Z.ai's GLM-5.3 in NVFP4 with DFlash2 speculative decoding: a 750B-parameter mixture-of-experts model for coding, agentic workflows, reasoning, tool use, and long-context tasks over a 1M-token window."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-glm-5-3: pricing · nebius-glm-5-3: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "zai-org/GLM-5.3-Flash" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/zai-org/GLM-5.3-Flash. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.15 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.50 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.03 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 153.8 (90% range 152.0–155.8) · Epoch AI score for GLM-5.3-Flash, model dated Aug 20, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-glm-5-3-flash: pricing · nebius-glm-5-3-flash: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "openai/gpt-oss-120b" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp4; weights repository https://huggingface.co/openai/gpt-oss-120b. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.15 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.60 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.10 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 140.2 (90% range 138.2–142.1) · range spans bands · Epoch AI score for gpt-oss-120b, model dated Aug 5, 2025
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-gpt-oss-120b: pricing · nebius-gpt-oss-120b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "NousResearch/Hermes-4-405B" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/NousResearch/Hermes-4-405B. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.00 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.20 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-hermes-4-405b: pricing · nebius-hermes-4-405b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "moonshotai/Kimi-K2.6" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization int4; weights repository https://huggingface.co/moonshotai/Kimi-K2.6. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.95 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.16 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists moonshotai/Kimi-K2.6 with recommended replacement moonshotai/Kimi-K3. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed moonshotai/Kimi-K2.6 from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-kimi-k2-6:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k2-6: pricing · nebius-kimi-k2-6: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists moonshotai/Kimi-K2.6 with recommended replacement moonshotai/Kimi-K3. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-kimi-k2-6:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed moonshotai/Kimi-K2.6 as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists moonshotai/Kimi-K2.6 again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "moonshotai/Kimi-K2.7-Code" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/moonshotai/Kimi-K2.7-Code. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.95 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.15 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists moonshotai/Kimi-K2.7-Code with recommended replacement moonshotai/Kimi-K3. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed moonshotai/Kimi-K2.7-Code from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-kimi-k2-7-code:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k2-7-code: pricing · nebius-kimi-k2-7-code: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists moonshotai/Kimi-K2.7-Code with recommended replacement moonshotai/Kimi-K3. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-kimi-k2-7-code:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed moonshotai/Kimi-K2.7-Code as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists moonshotai/Kimi-K2.7-Code again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Model ID: moonshotai/Kimi-K3 · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "moonshotai/Kimi-K3" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-west2; quantization fp4; weights repository https://huggingface.co/moonshotai/Kimi-K3. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $3.00 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $15.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.30 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 158.6 (90% range 156.6–160.8) · range spans bands · Epoch AI score for Kimi K3, model dated Jul 16, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k3: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "moonshotai/Kimi-K3", raw SHA-256 279526ca433255dc4e19562886346dcbc8c8d34f8bdfaf577b1713ad28135b17: "Moonshot AI's Kimi K3 frontier open-weights MoE model (NVFP4, 1M context) with DSpark speculative decoding, strong agentic tool use, reasoning, and coding."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k3: pricing · nebius-kimi-k3: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Model ID: moonshotai/Kimi-K3.1-Pro · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "moonshotai/Kimi-K3.1-Pro" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/moonshotai/Kimi-K3.1-Pro. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $2.00 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $10.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.20 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k3-1-pro: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "moonshotai/Kimi-K3.1-Pro", raw SHA-256 e97ed935250775ad0551e4474bed5119c3004fec0da2627825a01ffd2ee16c08: "Moonshot AI’s open-weight multimodal mixture-of-experts model for reasoning, coding and agentic tool use. It accepts text and images and generates text, with a 1M-token context window, 2.8T total parameters and 104B active parameters."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B. Labelled under the same rule after the offer's first publication (#1141); the description is identical on every archived day checked, 2026-11-05 to 2026-11-13.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-kimi-k3-1-pro: pricing · nebius-kimi-k3-1-pro: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Model ID: MiniMaxAI/MiniMax-M3 · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "MiniMaxAI/MiniMax-M3" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/MiniMaxAI/MiniMax-M3. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.30 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $1.20 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.06 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 147.7 (90% range 145.6–150.0) · range spans bands · Epoch AI score for MiniMax-M3, model dated Jun 1, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-minimax-m3: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "MiniMaxAI/MiniMax-M3", raw SHA-256 279526ca433255dc4e19562886346dcbc8c8d34f8bdfaf577b1713ad28135b17: "MiniMax-M3 is a 428B MoE reasoning model with 1M context, served on B200 via vLLM with EAGLE3 speculative decoding."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-minimax-m3: pricing · nebius-minimax-m3: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Model ID: MiniMaxAI/MiniMax-M3.1-Flash · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "MiniMaxAI/MiniMax-M3.1-Flash" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.15 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.60 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.015 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-minimax-m3-1-flash: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "MiniMaxAI/MiniMax-M3.1-Flash", raw SHA-256 e97ed935250775ad0551e4474bed5119c3004fec0da2627825a01ffd2ee16c08: "MiniMax-M3.1-Flash is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to 1M tokens."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B. Labelled under the same rule after the offer's first publication (#1141); the description is identical on every archived day checked, 2026-11-04 to 2026-11-13.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-minimax-m3-1-flash: pricing · nebius-minimax-m3-1-flash: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "meta-models/muse-spark-open" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B-NVFP4. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.25 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $4.25 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.15 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-muse-spark-open: pricing · nebius-muse-spark-open: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "nvidia/Nemotron-3.5-Super-VL" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-NVFP4. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.30 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.90 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.06 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-nemotron-3-5-super-vl: pricing · nebius-nemotron-3-5-super-vl: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Deprecated by provider · still listed · price checked Nov 18, 2026 (deprecation notice ↗)
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.06 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.24 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.012 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Superseded by the deprecation record of Nov 18, 2026, which availability now cites; kept with its original dates.
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B with recommended replacement nvidia/Nemotron-3_5-Lightning. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with no "deprecated" field, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the reviewAfter statement above, which describes the 2026-11-06 review: the notice, re-read at 2026-11-11T02:48:39Z (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1), states the same text and table. Nebius's models_info API still listed nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B as active on 2026-11-10 (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157), after the notice's removal date, with no "deprecated" field. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date. First reviewed 2026-11-06T16:33:57Z (#996).
Re-reviewed 2026-11-18 (#1098), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 and (#999) reviews: the notice, re-read at 2026-11-18T01:21:33Z (raw SHA-256 7189a6330d37f5f62b367ab067fce1128c1963e6412c5d379c06f4ecbe06674f), no longer lists nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B. Nebius's "Model Deprecations: December 2026" notice now lists it, removed from Serverless on December 24, 2026, as recorded by deprecation:nebius-nvidia-nemotron-3-nano-30b-a3b:2026-11-18, which availability now cites. This record keeps its retrieval date and its last confirming check (2026-11-11), and has no further review date.
Raw API evidence ↗ · provider API · Claims: nebius-nvidia-nemotron-3-nano-30b-a3b: pricing · nebius-nvidia-nemotron-3-nano-30b-a3b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: December 2026", raw SHA-256 3f0a861fb639cdfef57b0365700998b30bed664e076b17385b50afc449e17f6a (retrieved 2026-11-18T01:21:57Z): "On December 24, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on December 24, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B with recommended replacement nvidia/Nemotron-3_5-Lightning. It states: "Token Factory does not automatically reroute requests from deprecated models."
This replaces the November 2026 notice for this model: that notice, re-read at 2026-11-18T01:21:33Z (raw SHA-256 7189a6330d37f5f62b367ab067fce1128c1963e6412c5d379c06f4ecbe06674f), no longer lists nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B. Its removal text for November 9, 2026 and its other nine rows are unchanged from the 2026-11-11 copy (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1) that the #999 re-review of deprecation:nebius-nvidia-nemotron-3-nano-30b-a3b:2026-11-06 cites.
Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) listed nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B as active in every daily collection from 2026-11-12 to 2026-11-17; the latest (retrieved 2026-11-17T04:35:01.342Z, raw SHA-256 e97ed935250775ad0551e4474bed5119c3004fec0da2627825a01ffd2ee16c08, 34,283 bytes, 24 entries) is byte-identical to each collection since 2026-11-13. It lists the flavor with no "deprecated" field.
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the model is still listed as active and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #1098.
Deprecated by provider · still listed · price checked Nov 18, 2026 (deprecation notice ↗)
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "nvidia/nemotron-3-super-120b-a12b" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/nvidia/Nemotron-3-Super-120B. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.30 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.90 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.06 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Superseded by the deprecation record of Nov 18, 2026, which availability now cites; kept with its original dates.
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists nvidia/nemotron-3-super-120b-a12b with recommended replacement nvidia/Nemotron-3.5-Super-VL. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the reviewAfter statement above, which describes the 2026-11-06 review: the notice, re-read at 2026-11-11T02:48:39Z (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1), states the same text and table. Nebius's models_info API still listed nvidia/nemotron-3-super-120b-a12b as active on 2026-11-10 (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157), after the notice's removal date, with the flavor field "deprecated": false. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date. First reviewed 2026-11-06T16:33:57Z (#996).
Re-reviewed 2026-11-18 (#1098), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 and (#999) reviews: the notice, re-read at 2026-11-18T01:21:33Z (raw SHA-256 7189a6330d37f5f62b367ab067fce1128c1963e6412c5d379c06f4ecbe06674f), no longer lists nvidia/nemotron-3-super-120b-a12b. Nebius's "Model Deprecations: December 2026" notice now lists it, removed from Serverless on December 24, 2026, as recorded by deprecation:nebius-nemotron-3-super-120b-a12b:2026-11-18, which availability now cites. This record keeps its retrieval date and its last confirming check (2026-11-11), and has no further review date.
Raw API evidence ↗ · provider API · Claims: nebius-nemotron-3-super-120b-a12b: pricing · nebius-nemotron-3-super-120b-a12b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: December 2026", raw SHA-256 3f0a861fb639cdfef57b0365700998b30bed664e076b17385b50afc449e17f6a (retrieved 2026-11-18T01:21:57Z): "On December 24, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on December 24, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists nvidia/nemotron-3-super-120b-a12b with recommended replacement nvidia/Nemotron-3.5-Super-VL. It states: "Token Factory does not automatically reroute requests from deprecated models."
This replaces the November 2026 notice for this model: that notice, re-read at 2026-11-18T01:21:33Z (raw SHA-256 7189a6330d37f5f62b367ab067fce1128c1963e6412c5d379c06f4ecbe06674f), no longer lists nvidia/nemotron-3-super-120b-a12b. Its removal text for November 9, 2026 and its other nine rows are unchanged from the 2026-11-11 copy (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1) that the #999 re-review of deprecation:nebius-nemotron-3-super-120b-a12b:2026-11-06 cites.
Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) listed nvidia/nemotron-3-super-120b-a12b as active in every daily collection from 2026-11-12 to 2026-11-17; the latest (retrieved 2026-11-17T04:35:01.342Z, raw SHA-256 e97ed935250775ad0551e4474bed5119c3004fec0da2627825a01ffd2ee16c08, 34,283 bytes, 24 entries) is byte-identical to each collection since 2026-11-13. It lists the flavor with the flavor field "deprecated": false.
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the model is still listed as active and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #1098.
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "nvidia/Nemotron-3-Ultra-550b-a55b" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $1.00 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.12 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-nemotron-3-ultra-550b-a55b: pricing · nebius-nemotron-3-ultra-550b-a55b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "nvidia/Nemotron-3_5-Lightning" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization bf16; weights repository https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.06 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.24 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.03 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-nemotron-3-5-lightning: pricing · nebius-nemotron-3-5-lightning: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "openbmb/MiniCPM-V-4_5" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp16; weights repository https://huggingface.co/openbmb/MiniCPM-V-4_5. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.658 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $1.11 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.13 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists openbmb/MiniCPM-V-4_5 with recommended replacement nvidia/Nemotron-3.5-Super-VL. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": true. The flag is undocumented (#962) and is not the basis of this record; the dated notice is.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed openbmb/MiniCPM-V-4_5 from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-minicpm-v-4-5:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-minicpm-v-4-5: pricing · nebius-minicpm-v-4-5: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists openbmb/MiniCPM-V-4_5 with recommended replacement nvidia/Nemotron-3.5-Super-VL. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-minicpm-v-4-5:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed openbmb/MiniCPM-V-4_5 as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists openbmb/MiniCPM-V-4_5 again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Deprecated by provider · still listed · price checked Nov 18, 2026 (deprecation notice ↗)
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "Qwen/Qwen3-235B-A22B-Instruct-2507" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.20 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.60 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.05 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists Qwen/Qwen3-235B-A22B-Instruct-2507 with recommended replacement Qwen/Qwen3-30B-A3B-Instruct-2507. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": true. The flag is undocumented (#962) and is not the basis of this record; the dated notice is.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the reviewAfter statement above, which describes the 2026-11-06 review: the notice, re-read at 2026-11-11T02:48:39Z (raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1), states the same text and table. Nebius's models_info API still listed Qwen/Qwen3-235B-A22B-Instruct-2507 as active on 2026-11-10 (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157), after the notice's removal date, with the flavor field "deprecated": true. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date. First reviewed 2026-11-06T16:33:57Z (#996).
Re-reviewed 2026-11-18 (#1098), superseding the reviewAfter statement above, which describes the (#999) review: the notice, re-read at 2026-11-18T01:21:33Z (raw SHA-256 7189a6330d37f5f62b367ab067fce1128c1963e6412c5d379c06f4ecbe06674f), still lists Qwen/Qwen3-235B-A22B-Instruct-2507 with recommended replacement Qwen/Qwen3-30B-A3B-Instruct-2507, and states the same removal text for November 9, 2026; it no longer lists the two Nemotron models, which moved to a December 2026 notice. Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) listed Qwen/Qwen3-235B-A22B-Instruct-2507 as active in every daily collection from 2026-11-12 to 2026-11-17; the latest (retrieved 2026-11-17T04:35:01.342Z, raw SHA-256 e97ed935250775ad0551e4474bed5119c3004fec0da2627825a01ffd2ee16c08, 34,283 bytes, 24 entries) is byte-identical to each collection since 2026-11-13. It lists the flavor with the flavor field "deprecated": true. That latest listing is eight days after the notice's removal date. It stays deprecated and priced as listed; no retirement is recorded while it is listed. reviewAfter is the next re-review date.
Raw API evidence ↗ · provider API · Claims: nebius-qwen3-235b-a22b-instruct-2507: pricing · nebius-qwen3-235b-a22b-instruct-2507: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "Qwen/Qwen3-30B-A3B-Instruct-2507" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region eu-north1; quantization fp8; weights repository https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.10 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $0.30 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.10 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 135.2 (90% range 132.2–138.0) · Epoch AI score for Qwen3-30B-A3B-Instruct (Jul 2025), model dated Jul 29, 2025
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-qwen3-30b-a3b-instruct-2507: pricing · nebius-qwen3-30b-a3b-instruct-2507: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "Qwen/Qwen3.5-397B-A17B" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/Qwen/Qwen3.5-397B-A17B. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.60 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.60 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.12 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed: · Price freshness: stale
API availability: not verified · Retired: Nov 9, 2026
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de (retrieved 2026-11-06T16:31:55Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs."
The notice's "Deprecating Model" table lists Qwen/Qwen3.5-397B-A17B with recommended replacement Qwen/Qwen3.8-27B. It states: "Token Factory does not automatically reroute requests from deprecated models."
The published prices are Serverless list prices from the models_info API, the usage the notice covers; Dedicated Endpoints are out of scope.
Availability stays available: the models_info API still listed the model as active when this was reviewed (2026-11-06), and the removal date is in the future, so no retirement is recorded before it happens. reviewAfter is the removal date, a re-review date, not a retirement.
Nebius's models_info API (raw SHA-256 c8226b0f1921d48f469005af4d5d228a6602dacabb3793ec9e28662a480913b5, 2026-11-06 daily archive) lists this flavor with "deprecated": false, and as active. The flag is undocumented and disagrees with the notice for this model (#962); the dated notice is the basis of this record.
The same notice text is in the documentation's llms-full.txt (raw SHA-256 99cee4c2b34501f641aeacc38cf8e8fba45b7a975af5e8f1f82123ad028c1cc2, retrieved 2026-11-06T14:31:08Z).
Reviewed by Claude (owner delegation 2026-09-27), #996.
Re-reviewed 2026-11-11 (#999), superseding the availability and reviewAfter statements above, which describe the 2026-11-06 review: Nebius removed Qwen/Qwen3.5-397B-A17B from the models_info API between the 2026-11-09 and 2026-11-10 collections, as the notice dated; the retirement is recorded by retirement:nebius-qwen3-5-397b-a17b:2026-11-11, so this record has no further review date. First reviewed 2026-11-06T16:33:57Z (#996).
Raw API evidence ↗ · provider API · Claims: nebius-qwen3-5-397b-a17b: pricing · nebius-qwen3-5-397b-a17b: availability · 56 more claims about 28 other catalog entries · Retrieved · Checked · Review due · Freshness: stale · Confidence: B
Raw SHA-256: 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Nebius Token Factory documentation, "Model Deprecations: November 2026", raw SHA-256 cbd50601dd7ab41cabffedef0a6f60753587e35db103b78339097a8007582ab1 (retrieved 2026-11-11T02:48:39Z): "On November 9, 2026: Affected Serverless models will no longer be available via API or Playground. Dedicated Endpoints for these models will be unaffected by this change." and "These models will be removed from Token Factory Serverless on November 9, 2026. After that date, you will not be able to select them in the Playground or send API requests using their model IDs." Its "Deprecating Model" table lists Qwen/Qwen3.5-397B-A17B with recommended replacement Qwen/Qwen3.8-27B. The text is unchanged from the 2026-11-06 copy (raw SHA-256 9fd3f945fe2b9a357c243ec57203dd2b025bbbcc501bcaccaaabf88a9f3563de) that deprecation:nebius-qwen3-5-397b-a17b:2026-11-06 quotes.
Observed removal, in Nebius's models_info API (tokenfactory.nebius.com/api/public/models_info) as archived by the daily refresh: the 2026-11-09 collection (retrieved 2026-11-09T04:31:27.328Z, raw SHA-256 0f3c08ebef5a341541096adec8ac77252a6959aebf924f7fc0934f5feaecbae8, 44,234 bytes, 31 entries) listed Qwen/Qwen3.5-397B-A17B as active; the 2026-11-10 collection (retrieved 2026-11-10T04:29:59.019Z, raw SHA-256 75818db5c912d5249da37ada7349e0014cc66f996cd39b7c84bce7f746f1c157, 35,780 bytes, 25 entries) no longer lists it. The 6 entries that left are exactly the 6 notice models recorded as retired; none was added. The removal happened between those two collections.
retiredAt is the notice's date, November 9, 2026. The notice states no time or time zone, so it is stored as 2026-11-09T00:00:00Z and shown as the date only; the two collections above bracket the removal.
The rates kept are the last observed Serverless list prices (the 2026-11-08 record; the 2026-11-09 collection repeated its body byte for byte). Dedicated Endpoints, which the notice leaves unaffected, are out of scope. Availability is unknown, the delisting convention (#427/#433): no current listing remains to cite.
If Nebius lists Qwen/Qwen3.5-397B-A17B again, the collector fails for review (#999); a refresh never undoes or overwrites a retirement.
Reviewed by Claude (owner delegation 2026-09-27), #999.
Model ID: Qwen/Qwen3.8-27B · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "Qwen/Qwen3.8-27B" (the Nebius Token Factory API model ID), listed as an active image2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-central1; quantization fp4; weights repository https://huggingface.co/Qwen/Qwen3.8-27B. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.45 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $3.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.10 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Capability (Epoch AI): ECI 152.6 (90% range 150.6–154.8) · Epoch AI score for Qwen 3.8 27B, model dated Aug 14, 2026
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-qwen3-8-27b: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "Qwen/Qwen3.8-27B", raw SHA-256 c68067aa22da0c7c3a30ffe05c3aec831403e1813a80aa1ae77a0d38559cb575: "This NVFP4 serverless deployment supports tool calling and a 1M-token context window."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B. Replaces the 2026-10-17 record ("a native 262K context window"): the description changed on 2026-11-04 (#962).
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-qwen3-8-27b: pricing · nebius-qwen3-8-27b: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
Model ID: reflection/beam · Context window: 1M tokens
All rates, conditions and sources
Conditions of every rate below:
Nebius Token Factory public model catalog API (tokenfactory.nebius.com/api/public/models_info, which Nebius's model catalog calls "the authoritative machine-readable source"): input_price_per_million_tokens / output_price_per_million_tokens / input_cached_price_per_million_tokens, per 1 million tokens.
Model "reflection/beam" (the Nebius Token Factory API model ID), listed as an active text2text model with one flavor, labelled "cheap", and external_provider false.
As listed: serving region us-north1; quantization fp4; weights repository https://huggingface.co/reflection/Beam-1.0-501B-A23B. Hosts may serve different builds of a model; this listing is not proof of the developer's own weights.
Currency: the API names none; Nebius's Token Factory documentation states model-card prices "in USD per million tokens".
List prices: the API states no endpoint, account eligibility or regional price difference. It lists a cached-input price for this model; when cached input applies is not stated. Dedicated endpoints, post-training and embeddings are out of scope.
Input: $0.25 per 1M tokens · Standard tier
Input tokens (input_price_per_million_tokens).
Output: $1.00 per 1M tokens · Standard tier
Output tokens (output_price_per_million_tokens).
Cache read: $0.05 per 1M tokens · Standard tier
Cached input tokens (input_cached_price_per_million_tokens, which Nebius's model catalog lists among its token-model pricing fields). Nebius states no cache eligibility, TTL or cache-write price; they are unknown.
Price observed:
API availability: listed as available when checked (not a live service check)
Collection record
Raw API evidence ↗ · provider API · Claims: nebius-beam: specifications · Retrieved · Checked · Reviewed · Freshness: unknown · Confidence: B
Nebius Token Factory public model catalog API (models_info), model_description of the flavor "reflection/beam", raw SHA-256 279526ca433255dc4e19562886346dcbc8c8d34f8bdfaf577b1713ad28135b17: "It supports a 1M-token context window, was developed predominantly for English, and has a knowledge cutoff of June 30, 2026."
Shorthand stated in Nebius's own model description, recorded as stated under #504 (owner decision 2026-10-24, #622); not converted to an exact token count. Description prose, not a context field, hence confidence B.
Specifications only: no pricing, availability, capability or release-date assertion.
Raw API evidence ↗ · provider API · Claims: nebius-beam: pricing · nebius-beam: availability · 42 more claims about 21 other catalog entries · Retrieved · Checked · Review due · Confidence: B
Raw SHA-256: 2918fd707e714f5d03cefdbdc0fdab9382b1acd44d6abad8f635fac5a369e84d; collector nebius-models-info-json-1.
Hosted listing and pricing observation only: the reviewed token models of the public catalog; no checkpoint, developer, license, context-window, modality or release-date assertions. Like the other hosted providers, a listing is not proof of identical weights or behavior: some listings name a quantized or third-party build.
The API names no currency; US dollars rest on Nebius's Token Factory documentation (model cards show prices "in USD per million tokens"), hence confidence B. Availability rests on Nebius's own API listing the model as active (with external_provider false).
About capability scores (Epoch AI)
Capability scores: Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks' [online resource], epoch.ai/benchmarks. Accessed Nov 12, 2026. Licensed CC BY 4.0; values here are rounded to one decimal place and grouped into bands.
The Epoch Capabilities Index (ECI) combines many benchmarks on a scale where Claude 3.5 Sonnet is 130 and GPT-5 is 150. Bands, on the displayed one-decimal score: Frontier 160+, Strong 150–159.9, Capable 140–149.9, Entry below 140. Each score has a 90% range, listed with each model's rates and sources; a range that crosses a band boundary is flagged there. Models are not ranked: neighbouring scores often overlap.
A benchmark is not your workload. Epoch's scores mostly reflect runs at the highest reasoning effort, which use more output tokens than default settings. Scores are dated and can shift between Epoch releases. A score belongs to the developer's model; a host listing shows the same model's score, but hosts may serve different versions or quantizations.
How headline rates are chosen
Headline rates use the directory's shared rule: the rate a buyer meets by default, chosen by billing, tier and context dimensions and never by price. Context caps, the bracket starting at zero and short-context variants are labeled. Every price is a listed price, not a service guarantee; each offer's details say what its source states about availability. Batch, off-peak and other variants remain in the details; where no single rate is the default, the table shows a dash.
No currency conversion is performed. Missing cache prices are unknown, not zero. Equal units do not establish equivalent models or service conditions. See the methodology and source register.