Cloud / AI API pricing
Compare AI model token prices.
Standard paid-tier list prices normalized to US dollars per one million tokens, with cached input shown separately and every row linked to its vendor source.
Verified vendor pricing
API token prices
18 of 18 models
| Provider / model | Input | Cached input | Output | Context | Source |
|---|---|---|---|---|---|
| AnthropicClaude Fable 5Cached input is the cache-hit/refresh rate; cache writes cost more. US-only inference is 1.1×. | $10.00 | $1.00 | $50.00 | — | Vendor ↗ |
| AnthropicClaude Haiku 4.5Cached input is the cache-hit/refresh rate; cache writes cost more. | $1.00 | $0.10 | $5.00 | — | Vendor ↗ |
| AnthropicClaude Opus 5Cached input is the cache-hit/refresh rate; Fast mode and cache writes have separate prices. | $5.00 | $0.50 | $25.00 | — | Vendor ↗ |
| AnthropicClaude Sonnet 5Cached input is the cache-hit/refresh rate; US-only inference is 1.1×. | $2.00 | $0.20 | $10.00 | — | Vendor ↗ |
| GoogleGemini 2.5 Flash-Lite (text/image/video)Text, image, and video input rate; audio input costs more. | $0.10 | $0.01 | $0.40 | — | Vendor ↗ |
| GoogleGemini 2.5 Pro (≤200K prompt)Higher rates apply when the prompt exceeds 200K tokens; output includes thinking tokens. | $1.25 | $0.125 | $10.00 | — | Vendor ↗ |
| GoogleGemini 3.1 Pro Preview (≤200K prompt)Preview model; higher rates apply when the prompt exceeds 200K tokens. | $2.00 | $0.20 | $12.00 | — | Vendor ↗ |
| GoogleGemini 3.5 Flash-LiteStandard paid tier; output includes thinking tokens. | $0.30 | $0.03 | $2.50 | — | Vendor ↗ |
| GoogleGemini 3.7 FlashPromotional through December 31, 2026; output includes thinking tokens. | $0.75 | $0.075 | $3.75 | — | Vendor ↗ |
| Moonshot AIKimi K2.6International direct-API rate; reasoning and preserved historical reasoning consume billed tokens. | $0.95 | $0.16 | $4.00 | 262K | Vendor ↗ |
| Moonshot AIKimi K2.7 CodeInternational direct-API rate; thinking is always enabled and billed. | $0.95 | $0.19 | $4.00 | 262K | Vendor ↗ |
| Moonshot AIKimi K2.7 Code HighspeedHigher-throughput K2.7 Code variant; international direct-API rate. | $1.90 | $0.38 | $8.00 | 262K | Vendor ↗ |
| Moonshot AIKimi K3International direct-API rate; reasoning content is billed and applicable taxes are excluded. | $3.00 | $0.30 | $15.00 | 1M | Vendor ↗ |
| OpenAIGPT-5.4 mini | $0.75 | $0.075 | $4.50 | — | Vendor ↗ |
| OpenAIGPT-5.4 nano | $0.20 | $0.02 | $1.25 | — | Vendor ↗ |
| OpenAIGPT-5.6 LunaStandard short-context rate; cache writes and long-context requests have separate prices. | $0.20 | $0.02 | $1.20 | — | Vendor ↗ |
| OpenAIGPT-5.6 SolStandard short-context rate; cache writes and long-context requests have separate prices. | $5.00 | $0.50 | $30.00 | — | Vendor ↗ |
| OpenAIGPT-5.6 TerraStandard short-context rate; cache writes and long-context requests have separate prices. | $2.00 | $0.20 | $12.00 | — | Vendor ↗ |
Prices are normalized for comparison, while material pricing conditions remain in each record. Last source review: .
Visual comparison
Price distribution by model
Both axes use a logarithmic scale so low- and high-cost models remain visible.
Bar length represents median output-token price. Exact input and output medians are shown beside each provider.
Read before comparing
The same unit does not mean the same offer.
Output prices can include reasoning or thinking tokens, depending on the provider. Prompt length, input modality, cache writes, geography, and service tier can change a listed rate. The first note under each model preserves its most important condition.
See the methodology for normalization rules and the source register for canonical pricing pages.