GPQA Diamond: leaderboard, scores and methodology
Graduate-level questions in physics, chemistry and biology, written to resist search shortcuts.
Top 10
Bars are scaled to the highest observed score. The leader sets the pace; the spread below it shows how much separation exists in this benchmark.
Leaderboard
| # | Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|---|
| 1 | GPT-5.2 Reasoning | OpenAI | 1000000 | $20.98 | 87.5 |
| 2 | GPT-5.1 Max | OpenAI | 2000000 | $20.50 | 87.9 |
| 3 | Claude Opus 4.5 Pro | Anthropic | 200000 | $20.98 | 89.1 |
| 4 | GPT-5.1 Reasoning | OpenAI | 400000 | $20.27 | 88.3 |
| 5 | Claude Sonnet 4.5 Reasoning | Anthropic | 400000 | $19.48 | 87.7 |
| 6 | GPT-5.2 | OpenAI | 2000000 | $19.71 | 88.0 |
| 7 | o4 Turbo | OpenAI | 2000000 | $19.34 | 86.6 |
| 8 | Claude Opus 4.5 Max | Anthropic | 1000000 | $20.90 | 87.8 |
| 9 | Grok 4 Turbo | xAI | 1000000 | $19.21 | 86.7 |
| 10 | Claude Sonnet 4.5 Max | Anthropic | 1000000 | $19.90 | 87.2 |
| 11 | GPT-5.2 Turbo | OpenAI | 1000000 | $20.73 | 87.4 |
| 12 | o4 Reasoning | OpenAI | 1000000 | $19.69 | 87.5 |
| 13 | Gemini 3 Pro Reasoning | Google DeepMind | 200000 | $19.69 | 87.5 |
| 14 | Qwen3-Max Max | Alibaba Cloud | 400000 | $18.53 | 85.9 |
| 15 | Claude Sonnet 4.5 Turbo | Anthropic | 1000000 | $18.37 | 85.9 |
| 16 | GPT-5.1 Pro | OpenAI | 200000 | $19.77 | 87.8 |
| 17 | GPT-5.2 Max | OpenAI | 1000000 | $21.39 | 89.2 |
| 18 | GPT-5.2 Pro | OpenAI | 2000000 | $21.32 | 89.6 |
| 19 | Claude Sonnet 4.5 Pro | Anthropic | 1000000 | $18.73 | 86.7 |
| 20 | o3 Max | OpenAI | 2000000 | $17.82 | 84.8 |
| 21 | o4 Pro | OpenAI | 1000000 | $19.37 | 87.3 |
| 22 | Claude Opus 4.5 Reasoning | Anthropic | 2000000 | $20.91 | 88.2 |
| 23 | Claude Sonnet 4.5 | Anthropic | 400000 | $17.84 | 86.0 |
| 24 | Grok 4 | xAI | 200000 | $17.90 | 86.5 |
| 25 | GPT-5.1 | OpenAI | 2000000 | $19.01 | 86.0 |
| 26 | Qwen3-Max | Alibaba Cloud | 400000 | $17.37 | 84.4 |
| 27 | o4 | OpenAI | 2000000 | $17.74 | 85.4 |
| 28 | GPT-5.2 Vision | OpenAI | 1000000 | $19.03 | 87.4 |
| 29 | Grok 4 Max | xAI | 2000000 | $19.70 | 86.6 |
| 30 | Claude Opus 4.5 Turbo | Anthropic | 2000000 | $19.57 | 85.7 |
| 31 | Grok 4 Reasoning | xAI | 2000000 | $19.67 | 87.8 |
| 32 | Gemini 3 Flash Reasoning | Google DeepMind | 128000 | $17.05 | 83.7 |
| 33 | Gemini 3 Pro Max | Google DeepMind | 2000000 | $20.57 | 87.9 |
| 34 | Claude Opus 4.5 | Anthropic | 400000 | $19.42 | 87.7 |
| 35 | Gemini 2.5 Pro Turbo | Google DeepMind | 1000000 | $16.32 | 83.9 |
| 36 | Gemini 2.5 Pro Pro | Google DeepMind | 400000 | $17.16 | 83.6 |
| 37 | DeepSeek-R1 Instruct | DeepSeek | 1000000 | $5.90 | 83.1 |
| 38 | Gemini 3 Pro Pro | Google DeepMind | 200000 | $19.80 | 87.3 |
| 39 | Claude 3.7 Sonnet Reasoning | Anthropic | 128000 | $16.30 | 83.7 |
| 40 | GPT-5.1 Turbo | OpenAI | 2000000 | $19.19 | 86.1 |
| 41 | Grok 3 Pro | xAI | 400000 | $16.61 | 83.8 |
| 42 | GPT-4.1 Max | OpenAI | 400000 | $17.34 | 84.7 |
| 43 | Gemini 3 Pro Turbo | Google DeepMind | 400000 | $19.62 | 86.4 |
| 44 | DeepSeek-V3.2 Instruct | DeepSeek | 128000 | $5.91 | 83.7 |
| 45 | Gemini 3 Flash Max | Google DeepMind | 1000000 | $17.43 | 85.0 |
| 46 | Grok 3 Max | xAI | 2000000 | $17.66 | 85.9 |
| 47 | o4 Vision | OpenAI | 200000 | $17.01 | 84.5 |
| 48 | Claude Opus 4.5 Vision | Anthropic | 200000 | $17.70 | 85.9 |
| 49 | Claude 3.7 Sonnet Pro | Anthropic | 128000 | $16.07 | 83.3 |
| 50 | Gemini 3 Pro Vision | Google DeepMind | 200000 | $17.59 | 85.5 |
Methodology
GPQA Diamond is included as a reasoning benchmark with a typical score range of 0 to 100. The values in this index are deterministic modelled estimates generated for comparison at scale. They are not official leaderboard results and should not be quoted as vendor benchmarks.
Each model receives a score derived from its capability tier, release generation and benchmark difficulty. Scores are normalised to a 0–100 scale so that models from different providers can be placed side by side. The average shown is the arithmetic mean across all 600 scored models.
Read the leaderboard as a relative map. A five-point gap is meaningful at the top and less meaningful in the middle of the distribution. Pair the score with price and context data before choosing a model.
How to use this leaderboard
- Use the top 10 to shortlist candidates, not to declare a winner.
- Check whether the leader is open weights or proprietary before assuming you can deploy it.
- Compare the score against input and output price on each model page.
- Look at the model's scores on related benchmarks before trusting a single number.
FAQ
What is GPQA Diamond?
Graduate-level questions in physics, chemistry and biology, written to resist search shortcuts.
Who leads GPQA Diamond?
GPT-5.2 Reasoning from OpenAI leads this index with a score of 87.0, followed by GPT-5.1 Max at 86.4.
What is a good GPQA Diamond score?
The top quartile starts around 74.6. The index average is 65.6, so a score above 74.6 is a strong result relative to this dataset.
How are GPQA Diamond scores calculated?
The values are modelled estimates on a 0–100 scale, generated from capability tier, release generation and benchmark difficulty. They are not official leaderboard results.
Which models are included?
All 600 models in the index are scored on GPQA Diamond, across 40 providers and every model category.