Benchmarks · reasoning
GPQA Diamond (Epoch AI run)
Graduate-level science multiple choice (the diamond subset), run by Epoch AI. Accuracy. Data: Epoch AI, CC BY.
166 models · 239 results · data as of 2026-10-29 · epoch.ai ↗
Leaderboard
Gemini 4 Argon high
All results
239 rows| # | Model | Effort | Score | Harness |
|---|---|---|---|---|
| 1 | Gemini 4 Argon | high | 96.4% | Epoch AI |
| 2 | GPT-6 Astra | max | 95.8% | Epoch AI |
| 3 | Sonnet 5.5 | max | 95.6% | Epoch AI |
| 4 | Gemini 3.8 Flash | high | 95.4% | Epoch AI |
| 5 | GPT-6.1 Sol | max | 95.4% | Epoch AI |
| 6 | Gemini 3.7 Flash | high | 94.8% | Epoch AI |
| 7 | GPT-5.4 Pro | xhigh | 94.6% | Epoch AI |
| 8 | Gemini 3.1 Pro | high | 94.4% | Epoch AI |
| 9 | GPT-6 Sol | max | 94.3% | Epoch AI |
| 10 | GPT-6.1 Terra | max | 94.3% | Epoch AI |
| 11 | Gemini 3.6 Flash | high | 94.1% | Epoch AI |
| 12 | GPT-5.5 | xhigh | 94.0% | Epoch AI |
| 13 | Grok 4.6 | high | 94.0% | Epoch AI |
| 14 | GPT-5.5 Pro | xhigh | 93.9% | Epoch AI |
| 15 | GPT-5.6 Sol | max | 93.5% | Epoch AI |
| 16 | Grok 4.5 | high | 93.4% | Epoch AI |
| 17 | GPT-5.6 Terra | max | 93.3% | Epoch AI |
| 18 | GPT-5.4 | xhigh | 93.3% | Epoch AI |
| 19 | DeepSeek V4.1 Flash 1017 | max | 93.2% | Epoch AI |
| 20 | Grok 4.6 | xhigh | 93.2% | Epoch AI |
| 21 | Kimi K3 | max | 93.1% | Epoch AI |
| 22 | Opus 5 | max | 92.9% | Epoch AI |
| 23 | Gemini 3.5 Flash | high | 92.8% | Epoch AI |
| 24 | Grok 4.7 | xhigh | 92.7% | Epoch AI |
| 25 | Gemini 3 Pro | high | 92.6% | Epoch AI |
| 26 | Qwen 3.8 Max | xhigh | 92.3% | Epoch AI |
| 27 | Kimi K3 | high | 91.9% | Epoch AI |
| 28 | GLM 5.2 | max | 91.9% | Epoch AI |
| 29 | DeepSeek V4 Pro 0813 | max | 91.7% | Epoch AI |
| 30 | GPT-5.6 Luna | max | 91.6% | Epoch AI |