Best price-performance AI models
Capability per dollar: the Artificial Analysis intelligence index divided by the blended price (3:1 input/output) per 1M tokens. Higher is better value.
Updated Oct 21, 2026
Best value among capable models (index 45+)
Cheap small models win any pure ratio, so this list keeps only models strong enough for demanding work.
| # | Model | Index | Blended / 1M | Points per $ |
|---|---|---|---|---|
| 1 | MiMo-V2.6-Pro · Xiaomi | 46.3 | $0.544 | 85.1 |
| 2 | Muse Spark 1.3 · Meta | 48.1 | $2.00 | 24.1 |
| 3 | Grok 4.7 · xAI | 46.4 | $3.00 | 15.5 |
| 4 | Qwen3.8 Max (0902) · Qwen | 45.4 | $3.00 | 15.1 |
| 5 | Claude Sonnet 5.5 · Anthropic | 56 | $4.00 | 14.0 |
| 6 | Gemini 4 Argon · Google | 52.6 | $4.00 | 13.2 |
| 7 | GPT-6.1 Sol · OpenAI | 51.8 | $4.00 | 12.9 |
| 8 | GPT-6 Sol · OpenAI | 47.6 | $4.00 | 11.9 |
| 9 | Claude Opus 5.5 · Anthropic | 57.6 | $8.00 | 7.2 |
| 10 | GPT-5.6 Sol · OpenAI | 47 | $8.00 | 5.9 |
| 11 | Claude Opus 5 · Anthropic | 50.8 | $10.00 | 5.1 |
| 12 | Claude Fable 5.1 · Anthropic | 53.4 | $20.00 | 2.7 |
| 13 | GPT-6 Astra · OpenAI | 52.7 | $20.00 | 2.6 |
| 14 | Claude Fable 5 · Anthropic | 49.6 | $20.00 | 2.5 |
Best value overall
| # | Model | Index | Blended / 1M | Points per $ |
|---|---|---|---|---|
| 1 | GLM 5.3 Flash · Z.ai | 41.8 | $0.061 | 686.4 |
| 2 | Ling 3.0 Flash · inclusionAI | 20.1 | $0.032 | 638.1 |
| 3 | Ling 3.1 Flash · inclusionAI | 41.1 | $0.112 | 365.3 |
| 4 | gpt-oss-20b · OpenAI | 10 | $0.036 | 277.8 |
| 5 | Solar Mini 4 · Upstage | 24.1 | $0.088 | 275.4 |
| 6 | Ling 3.0 Flash VL · inclusionAI | 24.6 | $0.09 | 273.3 |
| 7 | Claude Haiku 5.5 · Anthropic | 43.4 | $0.20 | 217.0 |
| 8 | MiMo-V2.6-Flash · Xiaomi | 37.9 | $0.175 | 216.6 |
| 9 | DeepSeek V4 Flash 0423 · DeepSeek | 24.4 | $0.114 | 214.5 |
| 10 | GPT-6 Luna · OpenAI | 38.1 | $0.20 | 190.5 |
| 11 | DeepSeek V4.1 Flash · DeepSeek | 24.7 | $0.131 | 189.3 |
| 12 | Mercury 2.5 · Inception | 12.3 | $0.068 | 182.2 |
| 13 | MiMo-V2.5 · Xiaomi | 21.8 | $0.149 | 146.6 |
| 14 | Qwen3.5-9B · Qwen | 13.3 | $0.092 | 143.8 |
| 15 | Nemotron 3.5 Lightning · NVIDIA | 12.9 | $0.103 | 125.9 |
| 16 | Granite 4.0 Micro · IBM Granite | 5.1 | $0.041 | 125.2 |
| 17 | Llama 3.1 8B Instruct · Meta Llama | 6.9 | $0.058 | 120.0 |
| 18 | Gemma 4 26B A4B · Google | 16.7 | $0.143 | 117.2 |
| 19 | Mistral Small 3 · Mistral AI | 6.7 | $0.058 | 116.5 |
| 20 | Step 3.5 Flash · Stepfun | 17 | $0.15 | 113.3 |
| 21 | Gemma 3 4B · Google | 4.8 | $0.042 | 112.9 |
| 22 | Nemotron 3.5 Super VL · NVIDIA | 26.5 | $0.237 | 111.6 |
| 23 | GLM 5.3 · Z.ai | 44.8 | $0.409 | 109.5 |
| 24 | DeepSeek V4 Flash Vision Exp · DeepSeek | 34.8 | $0.323 | 107.6 |
| 25 | Solar Pro 4 · Upstage | 28.2 | $0.262 | 107.4 |
| 26 | Granite 4.2 8B · IBM Granite | 11.1 | $0.107 | 103.3 |
| 27 | GLM 4.7 Flash · Z.ai | 14.9 | $0.145 | 102.5 |
| 28 | Nova Micro 1.0 · Amazon | 5.9 | $0.061 | 96.3 |
| 29 | GPT-5 Nano · OpenAI | 13 | $0.138 | 94.5 |
| 30 | Qwen3.8 27B · Qwen | 33.7 | $0.368 | 91.7 |
| 31 | Hy3 preview · Tencent | 25.3 | $0.285 | 88.8 |
| 32 | MiMo-V2.6-Pro · Xiaomi | 46.3 | $0.544 | 85.1 |
| 33 | Nemotron 3 Nano 30B A3B · NVIDIA | 8.9 | $0.105 | 84.8 |
| 34 | GPT-5.6 Luna · OpenAI | 37.3 | $0.45 | 82.9 |
| 35 | Qwen3 Coder 30B A3B Instruct · Qwen | 9.6 | $0.123 | 78.4 |
| 36 | DeepSeek V4.1 Flash 1017 · DeepSeek | 39.5 | $0.525 | 75.2 |
| 37 | MiMo-V2.5-Pro · Xiaomi | 26 | $0.381 | 68.3 |
| 38 | DeepSeek V3.2 · DeepSeek | 21.5 | $0.315 | 68.3 |
| 39 | Llama 3.2 1B Instruct · Meta Llama | 4.8 | $0.071 | 68.1 |
| 40 | Phi 4 · Microsoft | 5.9 | $0.088 | 67.4 |
The intelligence index comes from Artificial Analysis via OpenRouter and covers only scored models. See the methodology.