LLM cost calculator
- Cheapest: Mistral Nemo
- $0.03
- Median
- $1.70
- Most expensive: o1-pro
- $600.00
Output price per million tokens · read October 21
Cheapest output, $ per million tokens
- Mistral Nemo $0.03
- Gemma 3 4B $0.05
- S Llama 3 8B Lunaris $0.05
- Ling 3.0 Flash $0.063
- Mistral Small 3 $0.08
- Llama 3.1 8B Instruct $0.08
- gpt-oss-20b $0.09
- N Nex-N2.5-Mini $0.10
Monthly API cost is requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000. At 10,000 requests a month with 2,000 input and 500 output tokens each, the cheapest scored model, Qwen3.5-Flash, costs $2.60 a month and the most expensive scored model, OpenAI GPT-4, $900.00.
Prices are OpenRouter list prices from October 21, 2026, built into this page. Nothing you type leaves your browser.
Your workload
Start from
- Claude Opus 5.5Anthropic92.2$180.00
- Claude Fable 5.1Anthropic89.1$450.00
- Claude Sonnet 5.5Anthropic88.9$90.00
- GPT-6 AstraOpenAI88.7$450.00
- GPT-6.1 SolOpenAI88.7$90.00
- Qwen3.8 FlashQwen87.4$5.35
- o3 ProOpenAI84.9$800.00
- DeepSeek V4.1 Flash 1017DeepSeek83.3$6.00
- Gemini 2.5 Pro Preview 06-05Google83.1$75.00
- GPT-6 SolOpenAI82.8$90.00
40 of 136 matching models shown.
| Model | Evidence | Index | Per request | Per month | Cost per solved task |
|---|---|---|---|---|---|
| Qwen3.5-FlashQwen | Thin, 1 benchmark | 6.2 | $0.00026 | $2.60 | $0.40 |
| DeepSeek V4 Flash 0423DeepSeek | Thin, 1 benchmark | 54.6 | $0.00027 | $2.73 | No run cost |
| Qwen3 Coder 30B A3B InstructQwen | Thin, 1 benchmark | 60.4 | $0.00028 | $2.80 | No run cost |
| Gemma 4 26B A4BGoogle | Thin, 1 benchmark | 22.8 | $0.00028 | $2.80 | No run cost |
| GPT-5 NanoOpenAI | Thin, 1 benchmark | 34.8 | $0.0003 | $3.00 | $0.109 |
| Qwen3 32BQwen | Thin, 1 benchmark | 40.0 | $0.0003 | $3.00 | No run cost |
| Gemma 3 27BGoogle | Thin, 1 benchmark | 4.9 | $0.00034 | $3.35 | No run cost |
| Llama 4 ScoutMeta Llama | Thin, 1 benchmark | 9.1 | $0.00035 | $3.50 | $1.276 |
| MiMo-V2.5Xiaomi | Thin, 1 benchmark | 34.0 | $0.00036 | $3.57 | $0.01 |
| DeepSeek V4.1 FlashDeepSeek | Partial, 2 benchmarks | 77.2 | $0.00036 | $3.61 | No run cost |
| GPT-4.1 NanoOpenAI | Thin, 1 benchmark | 8.9 | $0.0004 | $4.00 | No run cost |
| MiMo-V2.6-FlashXiaomi | Thin, 1 benchmark | 61.9 | $0.00042 | $4.20 | No run cost |
| Claude Haiku 5.5Anthropic | Partial, 2 benchmarks | 75.9 | $0.00045 | $4.50 | No run cost |
| GPT-6 LunaOpenAI | Partial, 2 benchmarks | 71.4 | $0.00045 | $4.50 | No run cost |
| Hy3Tencent | Thin, 1 benchmark | 43.9 | $0.00053 | $5.28 | No run cost |
| Qwen3.8 FlashQwen | Thin, 1 benchmark | 87.4 | $0.00053 | $5.35 | No run cost |
| GLM 5.3 FlashZ.ai | Partial, 2 benchmarks | 71.0 | $0.00055 | $5.50 | $0.02 |
| Nemotron 3.5 Super VLNVIDIA | Thin, 1 benchmark | 65.5 | $0.00055 | $5.50 | No run cost |
| DeepSeek V4.1 Flash 1017DeepSeek | Partial, 2 benchmarks | 83.3 | $0.0006 | $6.00 | No run cost |
| Gemma 4 31BGoogle | Partial, 2 benchmarks | 38.3 | $0.0006 | $6.00 | No run cost |
| GPT-4o-mini (2024-07-18)OpenAI | Thin, 1 benchmark | 3.6 | $0.0006 | $6.00 | No run cost |
| gpt-oss-120bOpenAI | Partial, 2 benchmarks | 34.5 | $0.0006 | $6.00 | $0.51 |
| Solar Pro 4Upstage | Thin, 1 benchmark | 24.4 | $0.0006 | $6.00 | No run cost |
| Hy3 previewTencent | Thin, 1 benchmark | 22.4 | $0.00066 | $6.60 | No run cost |
| Llama 4 MaverickMeta Llama | Partial, 2 benchmarks | 18.1 | $0.0007 | $7.01 | $1.491 |
| DeepSeek V3.2 ExpDeepSeek | Partial, 2 benchmarks | 55.0 | $0.00075 | $7.45 | No run cost |
| DeepSeek V4 Flash Vision ExpDeepSeek | Thin, 1 benchmark | 37.6 | $0.00075 | $7.55 | No run cost |
| Qwen3.8 27BQwen | Partial, 2 benchmarks | 73.6 | $0.00076 | $7.55 | $1.26 |
| DeepSeek V3.2DeepSeek | Partial, 2 benchmarks | 54.4 | $0.00077 | $7.70 | $0.99 |
| Qwen3.5-35B-A3BQwen | Thin, 1 benchmark | 7.7 | $0.00078 | $7.80 | No run cost |
| GLM 5.3Z.ai | Partial, 2 benchmarks | 73.1 | $0.00086 | $8.58 | $0.36 |
| Mercury 2Inception | Thin, 1 benchmark | 0.0 | $0.00088 | $8.75 | No run cost |
| Trinity Large ThinkingArcee AI | Thin, 1 benchmark | 5.4 | $0.0009 | $9.00 | No run cost |
| MiMo-V2.5-ProXiaomi | Thin, 1 benchmark | 40.0 | $0.00091 | $9.14 | $0.03 |
| GPT-5.6 LunaOpenAI | Partial, 2 benchmarks | 61.5 | $0.001 | $10.00 | $0.04 |
| MiniMax M2.5MiniMax | Partial, 2 benchmarks | 59.4 | $0.00102 | $10.15 | $0.62 |
| MiniMax M2MiniMax | Partial, 2 benchmarks | 45.3 | $0.00102 | $10.20 | $0.702 |
| DeepSeek V3DeepSeek | Thin, 1 benchmark | 70.2 | $0.00103 | $10.29 | No run cost |
| Beam 1.0 501B A23BReflection | Thin, 1 benchmark | 33.8 | $0.0011 | $11.00 | No run cost |
| Qwen3 Coder 480B A35BQwen | Partial, 2 benchmarks | 50.0 | $0.0011 | $11.00 | No run cost |
What this does not include
The formula uses list prices only. It ignores cached-input discounts, batch discounts, long-context surcharges and image or audio tokens, none of which are in our data. Cost per solved task is a different measure: it comes from published SWE-bench run costs and is known for 84 models.