Benchmark scores get restated when harnesses change or models are re-run. This page keeps the trail instead of quietly overwriting it.
No historical snapshots recorded yet — the current scores are shown below.
| # | Model | Score | Observed |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 64.4% | 2026-10-20 |
| 2 | Claude Opus 5 | 56.3% | 2026-10-19 |
| 3 | GPT-6 Astra | 54.7% | 2026-10-20 |
| 4 | GPT 5.6-Sol | 44.5% | 2026-10-19 |
| 5 | Gemini 3.1 Pro | 44.4% | 2026-10-19 |
| 6 | Kimi K3 | 43.5% | 2026-10-19 |
| 7 | DeepSeek V4-Pro 0813 | 42.7% | 2026-10-19 |
| 8 | DeepSeek V4.1-Flash 1017 | 42.6% | 2026-10-20 |
| 9 | GLM 5.3 | 42% | 2026-10-19 |
| 10 | DeepSeek V4-Flash 0731 | 37.8% | 2026-10-19 |
| 11 | DeepSeek V4.1-Flash 0910 | 36.8% | 2026-10-19 |
Nothing has been overwritten yet.