Open data: AI models
Compare published indexes, API pricing and context limits.
Published values, excluding taxes. A missing value is not zero. Plans may share resources or charge by usage; check their conditions. Third-party indexes are excluded from these downloads.
Most recent review:
AI models
| Product | Input price (USD / M tokens) | Output price (USD / M tokens) | Context (tokens) | Max output (tokens) | Product review date |
|---|---|---|---|---|---|
| GPT-6 Astra | 10 | 50 | 1,050,000 | 128,000 | |
| GPT-6.1 Sol | 2 | 10 | 1,050,000 | 128,000 | |
| Claude Fable 5.1 | 10 | 50 | 1,000,000 | 128,000 | |
| Claude Opus 5.5 | 4 | 20 | 1,000,000 | 128,000 | |
| Claude Opus 5 | 5 | 25 | 1,000,000 | 128,000 | |
| GPT-5.6 Sol | 4 | 20 | 1,050,000 | 128,000 | |
| GPT-5.6 Terra | 2 | 12 | 1,050,000 | 128,000 | |
| GPT-6.1 Terra | 1 | 5 | 1,050,000 | 128,000 | |
| Gemini 4 Argon | 2 | 10 | 1,048,576 | 1,048,576 | |
| Gemini 3.8 Flash | 0.75 | 3.75 | 1,048,576 | 65,536 | |
| Claude Sonnet 5.5 | 2 | 10 | 1,000,000 | 128,000 | |
| Claude Sonnet 5 | 2 | 10 | 1,000,000 | 128,000 | |
| Claude Haiku 5.5 | 0.1 | 0.5 | 1,000,000 | 128,000 | |
| Claude Haiku 4.5 | 1 | 5 | 200,000 | 64,000 | |
| GPT-5.6 Luna | 0.2 | 1.2 | 1,050,000 | 128,000 | |
| GPT-6 Luna | 0.1 | 0.5 | 1,050,000 | 128,000 | |
| Mistral Large 3 | — | — | 256,000 | — | |
| Mistral Medium 3.5 | — | — | 256,000 | — | |
| Mistral Small 4 | 0.15 | 0.6 | 256,000 | — | |
| Ministral 3 · 14B | 0.2 | 0.2 | 256,000 | — | |
| Codestral · 25.08 | 0.3 | 0.9 | 128,000 | — | |
| DeepSeek V4.1 Flash | 0.3 | 1.2 | 1,000,000 | 384,000 | |
| DeepSeek V4 Pro | 1.32 | 3.96 | 1,000,000 | 384,000 | |
| Muse Spark 1.3 | 1.25 | 4.25 | — | — | |
| Kimi K3 | 3 | 15 | 1,048,576 | — | |
| Kimi K3.1 Pro | 2 | 10 | 1,048,576 | — | |
| GLM-5.3 | 1.4 | 4.4 | — | — | |
| Qwen3.8 Max | 2 | 6 | — | — | |
| GLM-5.3 Flash | 0.15 | 0.5 | — | — | |
| Qwen3.8 Flash | 0.15 | 0.47 | — | — | |
| MiniMax M3 | 0.3 | 1.2 | — | — | |
| MiniMax M3.1 Flash | 0.15 | 0.6 | — | — | |
| Grok 4.7 | 2 | 6 | 500,000 | — | |
| Grok 4.6 | 2 | 6 | 500,000 | — | |
| Mistral Large 4 | 0.7 | 2.1 | 1,000,000 | — | |
| Gemini 3.5 Flash-Lite | 0.3 | 2.5 | 1,048,576 | 65,536 |
What each field means
input· USD / M tokens- Standard uncached input price for short-context requests.
output· USD / M tokens- Standard output token price; reasoning tokens may be billed.
context· tokens- Published limit: total window for OpenAI/Claude, input limit for Gemini. It doesn't measure effective recall.
maxOutput· tokens- Output limit documented by the vendor.
Conditions and sources
GPT-6 Astra
Above 272K input tokens, the entire request costs 2× input and 1.5× output. Access depends on provider eligibility.
GPT-6.1 Sol
Prompts over 272K input tokens cost 2× input and 1.5× output for the whole request.
Claude Fable 5.1
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
Claude Opus 5.5
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
Claude Opus 5
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
GPT-5.6 Sol
Published promotional rate through November 30, 2026. Above 272K input tokens, the entire request costs 2× input and 1.5× output.
GPT-5.6 Terra
Above 272K input tokens, the entire request costs 2× input and 1.5× output. Access depends on provider eligibility.
GPT-6.1 Terra
Above 272K input tokens, the entire request costs twice the input and cache rates and 1.5 times the output rate; Batch and Flex cost half the standard rates.
Gemini 4 Argon
Through November 30, 2026: $2 input and $10 output per million; from December 1: $4 and $20. Output includes thinking. Unified input/output window: 1,048,576 tokens. Default synchronous output: 65,536; the 1,048,576 maximum requires long decode and shares the window with input.
Gemini 3.8 Flash
Promotional price until Dec 31, 2026: afterwards, input $1.50 and output $7.50/M. Output includes thinking. The context shown is the input limit, not the total window.
Claude Sonnet 5.5
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
Claude Sonnet 5
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
Claude Haiku 5.5
Tiered rate by prompt size: up to 100,000 tokens, $0.10 input and $0.50 output per M; above that, $0.50 and $2.50. Caching, batch and tools have their own terms.
Claude Haiku 4.5
Base rate for the direct API. Caching, batch, tools and other providers have their own terms. The index doesn't determine the outcome for your use case.
GPT-5.6 Luna
Above 272K input tokens, the entire request costs 2× input and 1.5× output. Access depends on provider eligibility.
GPT-6 Luna
Prompts over 272K input tokens cost 2× input and 1.5× output for the whole request.
Mistral Large 3
Deprecated since October 31, 2026. The provider recommends Mistral Large 4 for new integrations and no longer publishes a current rate for this model in its standard table. The previous price is not reused as a current price.
Mistral Medium 3.5
Deprecated since October 31, 2026. The provider recommends Mistral Large 4 for new integrations and no longer publishes a current rate for this model in its standard table. The previous price is not reused as a current price.
Mistral Small 4
Price of the API hosted by Mistral, not the cost of running the weights. Context shown per the provider's abbreviated figure; max output and indexes not included. Pin the identifier to catch regressions.
Ministral 3 · 14B
Price of the API hosted by Mistral, not the cost of running the weights. Context shown per the provider's abbreviated figure; max output and indexes not included. Pin the identifier to catch regressions.
Codestral · 25.08
Price of the API hosted by Mistral, not the cost of running the weights. Context shown per the provider's abbreviated figure; max output and indexes not included. Pin the identifier to catch regressions.
DeepSeek V4.1 Flash
Shows PEAK uncached rates. Peak hours: Monday–Friday 01–04 and 06–10 UTC. Off-peak costs half. The 1M/384K figures preserve published notation; reasoning tokens consume the output budget. The old deepseek-v4-flash alias remains accepted but serves DeepSeek-V4.1-Flash-1017; use deepseek-flash in integrations.
DeepSeek V4 Pro
Shows the PEAK uncached rate. Peak hours: Monday to Friday 01–04 and 06–10 UTC. Off-peak costs half. The 1M/384K figures keep the published notation; reasoning tokens use up budget.
Muse Spark 1.3
Standard rate; cache reads cost $0.15 per million. Context and maximum output not included.
Kimi K3
Standard rate; cache reads cost $0.30 per million. Context of 1,048,576 tokens. Maximum output not included.
Kimi K3.1 Pro
Standard rates; cached input costs $0.20 per million. Context: 1,048,576 tokens. Maximum output not incorporated.
GLM-5.3
Standard rate; cache reads cost $0.26 per million. Context and maximum output not included.
Qwen3.8 Max
Singapore and US-scope rate; the other global regions cost $1.65 input and $4.951 output per million. Context and maximum output not included.
Qwen3.8 Flash
Published international rates. Context and maximum output not incorporated.
MiniMax M3
Up to 512K input tokens: $0.30 input and $1.20 output per million; above that: $0.60 and $2.40. Includes a permanent 50% discount. Context and maximum output not incorporated.
MiniMax M3.1 Flash
Standard rates. Context and maximum output not incorporated.
Grok 4.7
Short-context rate. The provider charges different prices above the 200K-token threshold; tools and search are billed separately. Max output not included.
Grok 4.6
Short-context rate. The provider charges different prices above the 200K-token threshold; tools and search are billed separately. Max output not included.
Mistral Large 4
Public preview, released October 6, 2026. Standard Mistral API rates: $0.70 input and $2.10 output per million; cache $0.07/M. Mistral Ecosystem Open Weight License, not assumed equivalent to Apache 2.0. Maximum output is omitted without a confirmed limit.
Gemini 3.5 Flash-Lite
Standard input rate: $0.30/M; output: $2.50/M, including thinking tokens. Announced destination for gemini-2.5-flash from December 1, 2026. Displayed context is the input limit; cache, storage and grounding have separate charges.
License and attribution
Credit Codifly and link to this page when reusing the data. CC BY 4.0
Compare optionsGuides to help you decide
Prices read from official sources, worked examples and criteria for choosing. Download the data