Value ranking
Best Value LLMs for Knowledge in October 2026
As of October 2026, Qwen3.7 Flash gives the most knowledge performance per dollar: 48.9 points at a blended $0.06 per million tokens. gpt-oss-120b is next.
Last verified
Category score per dollar of blended API price, among models in the top half of the knowledge ranking.
- Qwen3.7 Flash 888.4 pts/$
- gpt-oss-120b 603.0 pts/$
- Nemotron 3 Nano 30B A3B 437.0 pts/$
- Nemotron 3.5 Lightning 429.1 pts/$
- Gemma 4 26B A4B IT 427.8 pts/$
- Qwen3.5-9B 408.4 pts/$
- Granite 4.2 8B 357.6 pts/$
- GPT-6 Luna 285.2 pts/$
- Hy3 282.7 pts/$
- Step 3.5 Flash 264.1 pts/$
Full ranking
| # | Model | Score | Input $/M | Output $/M | Points per $ |
|---|---|---|---|---|---|
| 1 | Qwen3.7 Flash | 48.9 | $0.03 | $0.13 | 888.4 |
| 2 | gpt-oss-120b | 42.4 | $0.037 | $0.17 | 603.0 |
| 3 | Nemotron 3 Nano 30B A3B | 38.2 | $0.05 | $0.20 | 437.0 |
| 4 | Nemotron 3.5 Lightning | 37.5 | $0.05 | $0.20 | 429.1 |
| 5 | Gemma 4 26B A4B IT | 45.7 | $0.0675 | $0.23 | 427.8 |
| 6 | Qwen3.5-9B | 46.0 | $0.10 | $0.15 | 408.4 |
| 7 | Granite 4.2 8B | 38.4 | $0.06 | $0.25 | 357.6 |
| 8 | GPT-6 Luna | 57.0 | $0.10 | $0.50 | 285.2 |
| 9 | Hy3 | 40.8 | $0.0825 | $0.33 | 282.7 |
| 10 | Step 3.5 Flash | 39.6 | $0.10 | $0.30 | 264.1 |
| 11 | Claude Haiku 5.5 | 50.8 | $0.10 | $0.50 | 254.1 |
| 12 | Gemma 4 31B IT | 37.9 | $0.09 | $0.34 | 248.7 |
| 13 | Qwen3.5-Flash | 43.2 | $0.10 | $0.40 | 246.9 |
| 14 | GLM-5.3-Flash | 58.4 | $0.15 | $0.50 | 245.8 |
| 15 | MiMo-V2.6-Flash | 42.2 | $0.14 | $0.28 | 240.9 |
| 16 | MiMo-V2.5 | 40.8 | $0.14 | $0.28 | 233.3 |
| 17 | MiMo-V2-Omni | 40.5 | $0.14 | $0.28 | 231.4 |
| 18 | MiMo-V2-Flash | 39.7 | $0.14 | $0.28 | 227.0 |
| 19 | Nemotron 3 Super | 38.9 | $0.08 | $0.45 | 225.3 |
| 20 | DeepSeek V4.1 Flash | 57.9 | $0.15 | $0.60 | 220.5 |
| 21 | DeepSeek V4 Flash | 55.4 | $0.15 | $0.60 | 211.2 |
| 22 | Qwen3-30B-A3B | 41.8 | $0.12 | $0.50 | 194.4 |
| 23 | DeepSeek-V3.2-Exp | 51.7 | $0.26 | $0.38 | 178.1 |
| 24 | GPT-5.6 Luna | 58.5 | $0.20 | $1.20 | 130.1 |
| 25 | MiniMax-M3 | 58.4 | $0.30 | $1.20 | 111.3 |
| 26 | Trinity Large Thinking | 40.9 | $0.25 | $0.80 | 105.6 |
| 27 | DeepSeek-V3.1 | 43.7 | $0.25 | $0.95 | 102.8 |
| 28 | Qwen3.6 Flash | 42.1 | $0.19 | $1.13 | 99.9 |
| 29 | DeepSeek-V3 | 37.5 | $0.24 | $0.90 | 92.7 |
| 30 | Qwen3.6 35B-A3B | 51.3 | $0.25 | $1.49 | 92.1 |
| 31 | GPT-5.4 nano | 41.9 | $0.20 | $1.25 | 90.5 |
| 32 | GLM-4.6V | 38.0 | $0.30 | $0.90 | 84.5 |
| 33 | MiMo-V2.6-Pro | 43.5 | $0.43 | $0.87 | 80.0 |
| 34 | Qwen3.7 Plus | 54.9 | $0.40 | $1.60 | 78.5 |
| 35 | MiMo-V2.5-Pro | 42.2 | $0.43 | $0.87 | 77.7 |
| 36 | MiMo-V2-Pro | 41.4 | $0.43 | $0.87 | 76.2 |
| 37 | Solar Pro4 | 39.8 | $0.30 | $1.20 | 75.8 |
| 38 | Inkling-Small | 48.2 | $0.45 | $1.20 | 75.7 |
| 39 | MiniMax-M2.5 | 39.2 | $0.30 | $1.20 | 74.7 |
| 40 | Gemini 3.1 Flash Lite | 41.9 | $0.25 | $1.50 | 74.4 |
| 41 | MiniMax-M2.1 | 38.3 | $0.30 | $1.20 | 72.9 |
| 42 | MiniMax-M2.7 | 37.7 | $0.30 | $1.20 | 71.9 |
| 43 | Qwen3.5 35B-A3B | 47.8 | $0.25 | $2 | 69.6 |
| 44 | GPT-5 Mini | 45.6 | $0.25 | $2 | 66.4 |
| 45 | Qwen3 14B | 39.3 | $0.35 | $1.40 | 64.1 |
| 46 | DeepSeek V4 Pro | 59.5 | $0.66 | $1.98 | 60.1 |
| 47 | Kimi K2.5 | 53.6 | $0.45 | $2.25 | 59.6 |
| 48 | Gemini 3.5 Flash Lite | 50.1 | $0.30 | $2.50 | 58.9 |
| 49 | Nemotron 3 Ultra | 52.5 | $0.50 | $2.20 | 56.8 |
| 50 | Gemini 3 Flash Preview | 58.8 | $0.50 | $3 | 52.3 |
| 51 | Qwen3.5 Plus | 46.0 | $0.40 | $2.40 | 51.1 |
| 52 | Nova 2 Lite | 43.0 | $0.30 | $2.50 | 50.6 |
| 53 | Gemini 3.8 Flash | 74.8 | $0.75 | $3.75 | 49.9 |
| 54 | Qwen3.6 Plus | 56.1 | $0.50 | $3 | 49.8 |
| 55 | DeepSeek-R1 | 44.5 | $0.50 | $2.15 | 48.7 |
| 56 | Qwen3-Next 80B-A3B Instruct | 41.8 | $0.50 | $2 | 47.7 |
| 57 | GLM-4.7 | 47.0 | $0.60 | $2.20 | 47.0 |
| 58 | Gemini 3.7 Flash | 69.7 | $0.75 | $3.75 | 46.5 |
| 59 | Qwen3.5 27B | 38.0 | $0.30 | $2.40 | 46.0 |
| 60 | Gemini 3.6 Flash | 67.8 | $0.75 | $3.75 | 45.2 |
| 61 | GLM-4.5V | 37.5 | $0.60 | $1.80 | 41.6 |
| 62 | Qwen3 235B-A22B | 49.6 | $0.70 | $2.80 | 40.5 |
| 63 | GLM-4.6 | 40.2 | $0.60 | $2.20 | 40.2 |
| 64 | Qwen3.5 397B-A17B | 53.3 | $0.60 | $3.60 | 39.5 |
| 65 | Qwen3.6 27B | 52.4 | $0.60 | $3.60 | 38.8 |
| 66 | Qwen3.8 27B | 41.6 | $0.99 | $1.49 | 37.3 |
| 67 | Kimi K2 (Jul 2025) | 37.3 | $0.57 | $2.30 | 37.2 |
| 68 | Seed 2.0 Pro | 40.2 | $0.50 | $3 | 35.7 |
| 69 | Qwen3.5 122B-A10B | 38.8 | $0.40 | $3.20 | 35.2 |
| 70 | Grok 4.20 (Non-Reasoning) | 52.8 | $1.25 | $2.50 | 33.8 |
| 71 | GLM-5 | 52.3 | $1 | $3.20 | 33.7 |
| 72 | Grok 4.3 | 52.5 | $1.25 | $2.50 | 33.6 |
| 73 | Qwen3-VL 235B-A22B | 40.3 | $0.70 | $2.80 | 32.9 |
| 74 | Qwen3 32B | 40.0 | $0.70 | $2.80 | 32.7 |
| 75 | Kimi K2.6 | 54.0 | $0.95 | $4 | 31.5 |
| 76 | Kimi K2.7 Code | 53.5 | $0.95 | $4 | 31.3 |
| 77 | GPT-5.4 mini | 51.5 | $0.75 | $4.50 | 30.5 |
| 78 | Step 5 Preview | 41.2 | $1 | $2.70 | 28.9 |
| 79 | GLM-5.3 | 58.3 | $1.40 | $4.40 | 27.1 |
| 80 | Muse Spark 1.2 | 54.1 | $1.25 | $4.25 | 27.1 |
| 81 | GLM-5.2 | 57.1 | $1.40 | $4.40 | 26.5 |
| 82 | Muse Spark 1.1 | 53.1 | $1.25 | $4.25 | 26.5 |
| 83 | Grok 4.20 Multi-Agent | 40.4 | $1.25 | $2.50 | 25.8 |
| 84 | GLM-5.1 | 54.9 | $1.40 | $4.40 | 25.6 |
| 85 | o4-mini | 43.6 | $1.10 | $4.40 | 22.6 |
| 86 | Inkling | 55.1 | $1.87 | $4.68 | 21.4 |
| 87 | GLM-5V-Turbo | 40.6 | $1.20 | $4 | 21.4 |
| 88 | Muse Spark 1.3 | 42.6 | $1.25 | $4.25 | 21.3 |
| 89 | Grok 4.6 | 63.3 | $2 | $6 | 21.1 |
| 90 | Grok 4.7 | 62.8 | $2 | $6 | 20.9 |
| 91 | Grok 4.5 | 62.3 | $2 | $6 | 20.8 |
| 92 | Qwen3.8 Max | 61.7 | $2 | $6 | 20.6 |
| 93 | Qwen3 Max | 48.1 | $1.20 | $6 | 20.0 |
| 94 | o3-mini | 38.3 | $1.10 | $4.40 | 19.9 |
| 95 | Qwen3.6 Max Preview | 57.6 | $1.30 | $7.80 | 19.7 |
| 96 | Gemini 3.5 Flash | 66.3 | $1.50 | $9 | 19.6 |
| 97 | Claude Haiku 4.5 | 37.7 | $1 | $5 | 18.8 |
| 98 | GPT-6.1 Sol | 71.8 | $2 | $10 | 17.9 |
| 99 | Claude Sonnet 5.5 | 66.0 | $2 | $10 | 16.5 |
| 100 | GPT-5 | 56.6 | $1.25 | $10 | 16.5 |
| 101 | Qwen3.7 Max | 61.6 | $2.50 | $7.50 | 16.4 |
| 102 | Gemini 2.5 Pro | 56.0 | $1.25 | $10 | 16.3 |
| 103 | GPT-6 Sol | 64.8 | $2 | $10 | 16.2 |
| 104 | Gemini 3.1 Pro Preview | 71.8 | $2 | $12 | 16.0 |
| 105 | o3 | 54.6 | $2 | $8 | 15.6 |
| 106 | GPT-5.1 | 50.6 | $1.25 | $10 | 14.7 |
| 107 | Claude Sonnet 5 | 55.6 | $2 | $10 | 13.9 |
| 108 | GPT-5.6 Terra | 61.2 | $2 | $12 | 13.6 |
| 109 | Mistral Medium 3.5 | 40.0 | $1.50 | $7.50 | 13.3 |
| 110 | GPT-5.2 | 59.3 | $1.75 | $14 | 12.3 |
| 111 | GPT-5.4 | 65.3 | $2.50 | $15 | 11.6 |
| 112 | Kimi K3 | 63.2 | $3 | $15 | 10.5 |
| 113 | Claude Sonnet 4.6 | 51.7 | $3 | $15 | 8.6 |
| 114 | Claude Opus 5.5 | 66.4 | $4 | $20 | 8.3 |
| 115 | Claude Sonnet 4.5 | 48.4 | $3 | $15 | 8.1 |
| 116 | GPT-5.3 Chat | 38.8 | $1.75 | $14 | 8.1 |
| 117 | GPT-5.6 Sol | 64.3 | $4 | $20 | 8.0 |
| 118 | Claude Sonnet 4 | 41.8 | $3 | $15 | 7.0 |
| 119 | Claude Opus 5 | 66.8 | $5 | $25 | 6.7 |
| 120 | Claude Opus 4.7 | 62.6 | $5 | $25 | 6.3 |
| 121 | Claude Opus 4.6 | 61.9 | $5 | $25 | 6.2 |
| 122 | Claude Opus 4.8 | 61.3 | $5 | $25 | 6.1 |
| 123 | GPT-5.5 | 64.4 | $5 | $30 | 5.7 |
| 124 | Claude Opus 4.5 | 56.5 | $5 | $25 | 5.6 |
| 125 | GPT-6 Astra | 75.3 | $10 | $50 | 3.8 |
| 126 | Claude Fable 5.1 | 69.6 | $10 | $50 | 3.5 |
| 127 | Claude Fable 5 | 62.2 | $10 | $50 | 3.1 |
| 128 | o1 | 41.5 | $15 | $60 | 1.6 |
| 129 | Claude Opus 4 | 44.0 | $15 | $75 | 1.5 |
| 130 | Claude Opus 4.1 | 42.0 | $15 | $75 | 1.4 |
| 131 | GPT-5 Pro | 56.7 | $15 | $120 | 1.4 |
| 132 | GPT-5.4 Pro | 68.3 | $30 | $180 | 1.0 |
| 133 | GPT-5.5 Pro | 64.1 | $30 | $180 | 0.9 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Compare the leaders
Frequently asked questions
What is the most cost-effective AI model for knowledge?
As of October 2026, Qwen3.7 Flash gives the most knowledge performance per dollar: 48.9 points at a blended $0.06 per million tokens. gpt-oss-120b is next.
What are the top 5 in this ranking?
Qwen3.7 Flash, gpt-oss-120b, Nemotron 3 Nano 30B A3B, Nemotron 3.5 Lightning, Gemma 4 26B A4B IT, in that order, as of October 2026.
How is this list ranked?
By score per dollar: the model’s score divided by its blended API price (three parts input to one part output), among models in the top half by score.