Kaggle Open Benchmarks - GPQA Diamond (Zero-Shot): leaderboard
Zero-shot accuracy on the 198-question GPQA Diamond subset of graduate-level, Google-proof multiple-choice science questions (biology, chemistry, physics). Kaggle's standardized re-run on its own harness under the open-benchmarks organization, not the original authors' leaderboard; Kaggle's fraction is shown as a percentage (0-100, higher is better). Four answer options, so guessing scores about 25.
Metric: Accuracy (%). Source: www.kaggle.com. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.6 Flash | 93.43 | #57 |
| 2 | GPT-5.6 Sol | 92.93 | #22 |
| 3 | GPT-6 Astra | 92.93 | #6 |
| 4 | Gemini 3.5 Flash | 91.41 | #66 |
| 5 | Grok 4.5 | 90.91 | #54 |
| 6 | Claude Opus 5 | 90.86 | #16 |
| 7 | Gemini 3 Flash (Preview) | 90.4 | #90 |
| 8 | GPT-5.6 Terra | 86.36 | #40 |
| 9 | Gemma 4 31B | 84.85 | #212 |
| 10 | GPT-5 | 84.34 | #105 |
| 11 | GPT-5.6 Luna | 84.34 | #79 |
| 12 | Claude Opus 4.8 | 84.26 | #39 |
| 13 | Gemma 4 26B A4B | 83.25 | #360 |
| 14 | DeepSeek R1 0528 | 80.81 | #232 |
| 15 | Gemini 3.5 Flash Lite | 74.75 | #174 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=kaggle-open-benchmarks-gpqa-diamond-zero-shot · How It Works · Data refreshed daily, snapshot 2026-11-05.