Benchmarks / Reasoning
AutoResearchExam
Uses mean hidden-test AUARC across 24-hour research tasks. Score: hidden-test AUARC. Collected by Jake Boggs' benchmark dashboard.
AutoResearchExam
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5.5 | max | 60.9% |
| 2 | Fable 5.1 | max | 60.2% |
| 3 | GPT-6 Astra | max | 60.0% |
| 4 | Opus 5 | max | 57.9% |
| 5 | Grok 4.7 | max | 52.3% |
| 6 | GPT-5.6 Sol | max | 51.3% |
| 7 | Grok 4.6 | max | 49.3% |
| 8 | Gemini 3.8 Flash | max | 48.4% |
| 9 | Kimi K3 | max | 43.5% |
| 10 | Qwen 3.8 Max | max | 42.3% |
| 11 | Muse Spark 1.3 | max | 37.9% |