Fireworks AI API pricing

Input: USD 0.30 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.006 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 1.20 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.375 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.0075 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 1.50 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.45 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.009 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 1.80 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.5625 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.01125 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 2.25 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "DeepSeek V4.1 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/deepseek-v4p1-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 3.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.30 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 15.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 3.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.375 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 18.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Ember-1". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/ember-1; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 1.40 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.26 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 4.40 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 1.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.325 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 5.50 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 2.10 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Input tokens (the first price, "input").

Cache read: USD 0.39 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 6.60 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Output tokens (the third price, "output").

Input: USD 2.10 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.39 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 6.60 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 2.625 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.4875 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 8.25 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.15 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.03 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.50 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.1875 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.0375 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.625 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.225 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.045 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.75 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.28125 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.05625 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.9375 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "GLM 5.3 Flash (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/glm-5p3-flash; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 3.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.30 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 15.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 3.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.375 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 18.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 4.50 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Input tokens (the first price, "input").

Cache read: USD 0.45 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 22.50 / 1000000 tokens · Fast · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: fast · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• The Fast serverless mode, priced in the "Standard" column of the adjacent Fast row ("Fast variants appear as adjacent rows"); that row's Priority cell is "—", so the page offers no Priority price for Fast.
• Output tokens (the third price, "output").

Input: USD 4.50 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.45 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 22.50 / 1000000 tokens · Standard · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 5.625 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.5625 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 28.125 / 1000000 tokens · Priority · Modality: unspecified

Region: us · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Kimi K3 (US)". The page does not define the "(US)" label; region "us" is recorded from the row label only, with no data-residency, endpoint or eligibility claim. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/kimi-k3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.30 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.06 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 1.20 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.45 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.09 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 1.80 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "MiniMax M3". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/minimax-m3; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.60 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.12 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 2.40 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.75 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.15 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 3.00 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3 Ultra (Preview)". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-3-ultra-nvfp4; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.05 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.01 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.20 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.0625 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.0125 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.25 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "NVIDIA Nemotron 3.5 Lightning 30B A3B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/nemotron-lightning-3p5-30b-a3b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 0.15 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.015 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.60 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 0.18 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.018 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 0.72 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "OpenAI GPT OSS 120B". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/gpt-oss-120b; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").

Input: USD 2.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Input tokens (the first price, "input").

Cache read: USD 0.25 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 6.00 / 1000000 tokens · Standard · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: standard · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Standard" column: the Standard serverless mode.
• Output tokens (the third price, "output").

Input: USD 3.00 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Input tokens (the first price, "input").

Cache read: USD 0.375 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Cached input tokens served from prompt cache (the second price, "cached input"). The source states no cache-write price or cache TTL.

Output: USD 9.00 / 1000000 tokens · Priority · Modality: unspecified

Region: unknown · Endpoint scope: unknown · Tier: priority · Billing: Unknown · Prompt bracket: none stated · Cache TTL: Unknown seconds · Valid through: Unknown

• Fireworks AI serverless per-token price from the "Text and vision models" table, USD per 1M tokens ("Prices below are per 1 million tokens in US dollars"). Batch inference (the page states it "is billed at 50% of serverless pricing on both input and output"), the size-based "Other base models" prices, embeddings, on-demand GPU deployments and training are out of scope.
• Price row "Qwen 3.8 Max". The page states no region or endpoint for this row. The row links the Fireworks model page https://app.fireworks.ai/models/fireworks/qwen3p8-max; the pricing page states no API model ID.
• "Priority" column: the Priority serverless mode, priced on the same row. The source states: "— in the Priority column means Priority is not available for that model. This pricing table is the source of truth for Priority availability."
• Output tokens (the third price, "output").