The cost of tokens across models on OpenRouter, ranked cheapest-first by input price — the falling cost of intelligence, measured daily. Real daily usage is shown where the model appears in usage data.
USD per million tokens, cheapest first. Priced models only — free tiers and unpriced variants excluded. Tokens/day shown where available. Trend is input price over time.
| # | Name | $/Mtok in | $/Mtok out | Context | Tokens/day | Price trend |
|---|---|---|---|---|---|---|
| 1 | inclusionAI: Ling-2.6-flash | 0.01 | 0.03 | 262.1k | — | |
| 2 | IBM: Granite 4.0 Micro | 0.017 | 0.112 | 131.0k | — | |
| 3 | mistralai/mistral-nemo | 0.019000000000000003 | 0.03 | 131.1k | 41.6B | |
| 4 | Nex AGI: Nex-N2-Mini | 0.024999999999999998 | 0.09999999999999999 | 262.1k | — | |
| 5 | OpenAI: GPT-5 Nano (batch) | 0.024999999999999998 | 0.19999999999999998 | 400.0k | — | |
| 6 | Meta: Llama 3.2 1B Instruct | 0.027 | 0.201 | 60.0k | — | |
| 7 | openai/gpt-oss-20b | 0.03 | 0.13 | 131.1k | 17.9B | |
| 8 | Qwen: Qwen3.7 Flash | 0.03 | 0.13 | 1.0M | — | |
| 9 | Amazon: Nova Micro 1.0 | 0.035 | 0.14 | 128.0k | — | |
| 10 | openai/gpt-oss-120b | 0.037 | 0.16999999999999998 | 131.1k | 78.2B | |
| 11 | Cohere: Command R7B (12-2024) | 0.0375 | 0.15 | 128.0k | — | |
| 12 | Sao10K: Llama 3 8B Lunaris | 0.04 | 0.049999999999999996 | 8.2k | — | |
| 13 | Qwen: Qwen3 30B A3B Instruct 2507 | 0.04815 | 0.19305 | 262.1k | — | |
| 14 | IBM: Granite 4.1 8B | 0.049999999999999996 | 0.09999999999999999 | 131.1k | — | |
| 15 | NVIDIA: Nemotron 3 Nano 30B A3B | 0.049999999999999996 | 0.19999999999999998 | 262.1k | — | |
| 16 | OpenAI: GPT-5 Nano | 0.049999999999999996 | 0.39999999999999997 | 400.0k | — | |
| 17 | Google: Gemma 3 4B | 0.049999999999999996 | 0.09999999999999999 | 131.1k | — | |
| 18 | Google: Gemma 3 12B | 0.049999999999999996 | 0.15 | 131.1k | — | |
| 19 | Mistral: Mistral Small 3 | 0.049999999999999996 | 0.08 | 32.8k | — | |
| 20 | Meta: Llama 3.2 3B Instruct | 0.049999999999999996 | 0.33 | 131.1k | — | |
| 21 | meta-llama/llama-3.1-8b-instruct | 0.049999999999999996 | 0.08 | 131.1k | 11.3B | |
| 22 | Google: Gemini 2.5 Flash Lite (batch) | 0.049999999999999996 | 0.19999999999999998 | 1.0M | — | |
| 23 | Poolside: Laguna XS 2.1 | 0.06 | 0.12 | 262.1k | — | |
| 24 | Z.ai: GLM 4.7 Flash | 0.06 | 0.39999999999999997 | 202.8k | — | |
| 25 | Google: Gemma 3n 4B | 0.06 | 0.12 | 32.8k | — | |
| 26 | Amazon: Nova Lite 1.0 | 0.06 | 0.24 | 300.0k | — | |
| 27 | MythoMax 13B | 0.06 | 0.06 | 8.2k | — | |
| 28 | Tencent: Hy3 preview | 0.063 | 0.21 | 262.1k | — | |
| 29 | Qwen: Qwen3.5-Flash | 0.065 | 0.26 | 1.0M | — | |
| 30 | Google: Gemma 4 26B A4B | 0.07 | 0.33999999999999997 | 262.1k | — | |
| 31 | Qwen: Qwen3 Coder 30B A3B Instruct | 0.07 | 0.28 | 262.1k | — | |
| 32 | Microsoft: Phi 4 | 0.07 | 0.14 | 16.4k | — | |
| 33 | inclusionAI: Ring-2.6-1T | 0.075 | 0.625 | 262.1k | — | |
| 34 | inclusionAI: Ling-2.6-1T | 0.075 | 0.625 | 262.1k | — | |
| 35 | ByteDance Seed: Seed 1.6 Flash | 0.075 | 0.3 | 262.1k | — | |
| 36 | OpenAI: gpt-oss-safeguard-20b | 0.075 | 0.3 | 131.1k | — | |
| 37 | Mistral: Mistral Small 3.2 24B | 0.075 | 0.19999999999999998 | 256.0k | — | |
| 38 | Qwen: Qwen3 32B | 0.08 | 0.28 | 131.1k | — | |
| 39 | Google: Gemma 3 27B | 0.08 | 0.44999999999999996 | 262.1k | 9.4B | |
| 40 | NVIDIA: Nemotron 3 Super | 0.08499999999999999 | 0.39999999999999997 | 1.0M | — |