GPQA Diamond AI Model Rankings
| # | Model | GPQA | Cost |
|---|---|---|---|
| 1 |  GPT 5.6 Sol OpenAI | 94.6% | $11.25/M |
| 1 | GPT 5.4 Pro OpenAI | 94.6% | $67.50/M |
| 2 |  Gemini 3.1 Pro Google | 94.1% | $4.50/M |
| 3 | GPT 5.5 OpenAI | 94.0% | $11.25/M |
| 4 | GPT 5.5 Pro OpenAI | 93.9% | $67.50/M |
| 5 | Kimi K3 MoonshotAI | 93.5% | $6.00/M |
| 6 |  Grok 4.5 xAI | 93.4% | $3.00/M |
| 7 | GPT 5.4 OpenAI | 93.3% | $5.63/M |
| 8 | GPT 5.6 Terra OpenAI | 92.9% | $5.63/M |
| 9 | Gemini 3.5 Flash Google | 92.8% | $3.38/M |
| 10 | Gemini 3 Pro Google | 92.6% | — |
| 11 | GPT 5.6 Luna OpenAI | 92.3% | $2.25/M |
| 12 |  GLM 5.2 Zhipu AI | 91.9% | $1.19/M |
| 13 |  Qwen3.7 Max Qwen | 91.6% | $2.21/M |
| 14 | GPT 5.2 OpenAI | 91.4% | $4.81/M |
| 15 |  Claude Opus 4.8 Anthropic | 91.0% | $10.00/M |
| 16 | Kimi K2.6 MoonshotAI | 90.8% | $1.37/M |
| 17 | Claude Sonnet 5 Anthropic | 90.5% | $4.00/M |
| 18 | Claude Opus 4.7 Anthropic | 90.2% | $10.00/M |
| 19 |  DeepSeek V4 Pro DeepSeek | 89.6% | $0.54/M |
| 20 | Kimi K2.7 Code MoonshotAI | 89.5% | $1.55/M |
| 21 | Grok 4.3 xAI | 88.8% | $1.56/M |
| 22 | GLM 5 Zhipu AI | 87.8% | $1.35/M |
| 23 | GPT 5.1 OpenAI | 87.6% | $3.44/M |
| 24 | Qwen3.6 Plus Qwen | 87.4% | $0.73/M |
| 25 |  Inkling Thinking Machines Lab | 87.2% | $1.76/M |
| 26 | GPT 5 OpenAI | 86.2% | $3.44/M |
| 27 | GLM 5.1 Zhipu AI | 85.5% | $1.48/M |
| 28 | Gemini 2.5 Pro Google | 85.3% | $3.44/M |
| 29 | GPT 5.4 Mini OpenAI | 83.6% | $1.69/M |
| 30 | GLM 4.7 Zhipu AI | 83.3% | $0.74/M |
| 31 | Gemini 3 Flash Google | 83.2% | $1.13/M |
| 32 | o3 OpenAI | 81.8% | $3.50/M |
| 33 | Claude Opus 4.5 Anthropic | 80.7% | $10.00/M |
| 34 | o4-mini (High) OpenAI | 79.6% | $1.93/M |
| 35 | GPT 5.4 Nano OpenAI | 78.5% | $0.46/M |
| 36 | o3-mini (High) OpenAI | 77.0% | $1.93/M |
| 37 | o1 OpenAI | 76.8% | $26.25/M |
| 38 | DeepSeek R1 (0528) DeepSeek | 76.3% | $0.91/M |
| 39 | GPT 5 Mini OpenAI | 75.0% | $0.69/M |
| 40 | Claude Sonnet 4.5 Anthropic | 73.7% | $6.00/M |
| 41 | Claude Opus 4.1 Anthropic | 73.2% | $30.00/M |
| 42 | Qwen3 Max Thinking Qwen | 72.6% | $1.56/M |
| 43 | Qwen3 235B A22B Qwen | 70.7% | $0.80/M |
| 44 | GPT 5 Nano OpenAI | 69.4% | $0.14/M |
| 45 | DeepSeek R1 DeepSeek | 69.2% | $1.15/M |
| 46 | Claude Opus 4 Anthropic | 69.2% | $30.00/M |
| 47 | DeepSeek V3 (0324) DeepSeek | 67.6% | $0.48/M |
| 48 | GPT 4.1 OpenAI | 66.9% | $3.50/M |
| 49 | Claude Sonnet 4 Anthropic | 66.7% | $6.00/M |
| 49 | Gemini 2.5 Pro (05-06) Google | 66.7% | $3.44/M |
| 50 | GPT 4.1 Mini OpenAI | 65.8% | $0.70/M |
| 51 | Claude Haiku 4.5 Anthropic | 60.5% | $2.00/M |
| 52 | DeepSeek V3 DeepSeek | 56.5% | $0.35/M |
| 53 | Qwen Max Qwen | 56.1% | — |
| 54 |  Phi 4 Microsoft | 56.1% | $0.09/M |
| 55 | DeepSeek R1 Distill Llama 70B DeepSeek | 55.7% | $0.80/M |
| 56 | Claude 3.5 Sonnet Anthropic | 55.3% | — |
| 57 |  Mistral Large 2 (Nov 2024) Mistral | 51.3% | — |
| 58 |  Llama 3.1 405B Meta | 50.9% | — |
| 59 | GPT 4o OpenAI | 49.2% | $4.38/M |
| 60 | Qwen2.5 72B Qwen | 49.1% | $0.37/M |
| 61 | Mistral Large 2 (Jul 2024) Mistral | 49.0% | $3.00/M |
| 62 | GPT 4.1 Nano OpenAI | 48.9% | $0.17/M |
| 63 | Gemma 3 27B Google | 48.9% | $0.15/M |
| 64 | Qwen Plus Qwen | 48.1% | $0.39/M |
| 65 | Llama 3.3 70B Meta | 47.4% | $0.20/M |
| 66 | GPT 4 Turbo OpenAI | 46.6% | $15.00/M |
| 67 | Llama 3.1 70B Meta | 44.2% | $0.40/M |
| 68 | WizardLM 2 8x22B Microsoft | 43.4% | $0.62/M |
| 69 | Qwen Turbo Qwen | 41.8% | — |
| 70 | Claude 3.5 Haiku Anthropic | 38.1% | $1.60/M |
| 71 | GPT 4o Mini OpenAI | 37.7% | $0.26/M |
| 72 | Gemma 2 27B Google | 36.5% | $0.65/M |
| 73 | Claude 3 Haiku Anthropic | 36.3% | $0.50/M |
| 74 | Gemma 2 9B Google | 27.5% | — |
| 75 | Llama 3.1 8B Meta | 25.9% | $0.06/M |
| 76 | Mistral 7B v0.3 Mistral | 15.2% | — |
78 of 78 models, ranked by GPQA Diamond