The Cheapest LLMs With a 1M-Token Context Window
Which million-token model costs least to run?
All 50 buyable models that hold 1M tokens or more, cheapest first. 31 of them also carry a published pass-rate score, shown alongside; the rest are listed without one rather than scored as zero.
Method: Every eligible model with a context window of 1M tokens or more, ordered by blended price per 1M tokens ascending (3:1 input-to-output tokens: 0.75 x input + 0.25 x output), cheapest first. The threshold is a published spec, not an editorial line, and no benchmark is required to appear, so nothing here is estimated. Pricing as of July 2026. 14 rows below carry a score measured at a non-default reasoning tier and are marked as such; every affected model is named on the hub. Read the full method.
| Rank | Model | Blended $/1M | Context | GPQA Diamond | Quality |
|---|---|---|---|---|---|
| #1 | Llama 4 ScoutMeta | $0.135 | 1.05M | 58.7 | 3.7 |
| #2 | DeepSeek V4-FlashDeepSeek· non-default tier | $0.175 | 1M | 86.7 | 56.9 |
| #3 | Gemini 2.5 Flash-LiteGoogle· non-default tier | $0.175 | 1.05M | 62.5 | not scored |
| #4 | GPT-4.1 NanoOpenAI | $0.175 | 1.05M | 51.2 | 3.7 |
| #5 | Grok 4.1 FastxAI· non-default tier | $0.275 | 2M | 85.3 | not scored |
| #6 | Grok 4.1 Fast ReasoningxAI· non-default tier | $0.275 | 2M | 85.3 | not scored |
| #7 | Llama 4 MaverickMeta | $0.415 | 1.05M | 67.1 | 7.9 |
| #8 | Nemotron 3 Super 120BNVIDIA | $0.425 | 1M | not published | not scored |
| #9 | DeepSeek V4-ProDeepSeek· non-default tier | $0.544 | 1M | 90.5 | 64.8 |
| #10 | MiMo-V2.5-ProXiaomi | $0.544 | 1.05M | 86.6 | 65.2 |
| #11 | Gemini 3.1 Flash-LiteGoogle | $0.563 | 1.05M | 82.2 | 31.1 |
| #12 | Qwen3.6-PlusAlibaba | $0.620 | 1M | 88.2 | 61.4 |
| #13 | GPT-4.1 MiniOpenAI | $0.700 | 1.05M | 66.4 | 10.1 |
| #14 | Qwen3.7 PlusAlibaba | $0.700 | 1M | 90.0 | 61.0 |
| #15 | Gemini 2.5 FlashGoogle· non-default tier | $0.850 | 1.05M | 79.0 | not scored |
| #16 | Gemini 3.5 Flash-LiteGoogle | $0.850 | 1.05M | 83.8 | 53.6 |
| #17 | Nova 2.0 LiteAmazon | $0.850 | 1M | 76.8 | not scored |
| #18 | Nemotron 3 Ultra 550BNVIDIA | $1.00 | 1M | not published | 71.9 |
| #19 | MiniMax M3MiniMax | $1.05 | 1M | 92.9 | 65.2 |
| #20 | Gemini 3 FlashGoogle | $1.13 | 1.05M | not published | not scored |
| #21 | Gemini 3 Flash ReasoningGoogle· non-default tier | $1.13 | 1.05M | 89.8 | not scored |
| #22 | LongCat-2.0Meituan | $1.30 | 1M | 78.0 | 50.2 |
| #23 | MiMo-V2-ProXiaomi | $1.50 | 1.05M | 87.0 | 78.0 |
| #24 | Grok 4.3xAI | $1.56 | 1M | 89.0 | not scored |
| #25 | Gemini 3.1 Flash LiveGoogle | $1.69 | 1.05M | not published | not scored |
| #26 | Muse Spark 1.1Meta | $2.00 | 1M | 89.8 | 77.9 |
| #27 | GLM-5.2Zhipu· non-default tier | $2.15 | 1M | 89.5 | 77.9 |
| #28 | GPT-5.6 LunaOpenAI | $2.25 | 1.05M | 85.9 | 53.2 |
| #29 | Gemini 3.6 FlashGoogle | $3.00 | 1.05M | 92.8 | 77.5 |
| #30 | Grok 4.20xAI· non-default tier | $3.00 | 2M | 91.1 | not scored |
| #31 | Gemini 3.5 FlashGoogle | $3.38 | 1.05M | 92.1 | 76.2 |
| #32 | Gemini 2.5 ProGoogle | $3.44 | 1.05M | 84.4 | 28.5 |
| #33 | GPT-4.1OpenAI | $3.50 | 1.05M | 66.6 | not scored |
| #34 | Qwen3.7 MaxAlibaba | $3.75 | 1M | 92.3 | 74.5 |
| #35 | Claude Sonnet 5Anthropic· non-default tier | $4.00 | 1M | 91.1 | 83.3 |
| #36 | Gemini 3 ProGoogle | $4.50 | 1.05M | not published | not scored |
| #37 | Gemini 3.1 ProGoogle | $4.50 | 1.05M | not published | not scored |
| #38 | GPT-5.4OpenAI· non-default tier | $5.63 | 1.05M | 87.1 | not scored |
| #39 | GPT-5.6 TerraOpenAI | $5.63 | 1.05M | 87.2 | 72.3 |
| #40 | Claude Sonnet 4.6Anthropic· non-default tier | $6.00 | 1M | 87.5 | 71.2 |
| #41 | Grok 4xAI | $6.00 | 2M | 87.7 | not scored |
| #42 | Kimi K3Moonshot | $6.00 | 1.05M | 93.5 | 85.0 |
| #43 | Claude Opus 4.6Anthropic· non-default tier | $10.00 | 1M | 89.6 | not scored |
| #44 | Claude Opus 4.7Anthropic· non-default tier | $10.00 | 1M | 91.4 | 83.1 |
| #45 | Claude Opus 4.8Anthropic | $10.00 | 1M | 92.0 | 87.0 |
| #46 | GPT-5.5OpenAI | $11.25 | 1.05M | 92.6 | 80.5 |
| #47 | GPT-5.6 SolOpenAI | $11.25 | 1.05M | 92.6 | 86.1 |
| #48 | Claude Fable 5Anthropic | $20.00 | 1M | 92.6 | 84.6 |
| #49 | GPT-5.4 ProOpenAI | $67.50 | 1.05M | not published | not scored |
| #50 | GPT-5.5 ProOpenAI | $67.50 | 1.05M | not published | not scored |
Frequently asked
Nearby questions, answered elsewhere
Ranked on window size instead
Every model by how much it holds, not what it costs. Same models at the top of the catalogue, a completely different order.
Cheapest models overall
The whole catalogue on a plain 1:1 input/output average with no context requirement at all. This page requires 1M tokens and blends prices 3:1.