CMT-Benchmark - reported pass@8
Science
Condensed Matter Theory benchmark pass@8 reported in Google's February 2026 Deep Think evaluation, with answer correctness judged by Gemini. Eight attempts are combined as pass@8, not averaged as eight pass@1 runs. The exact evaluated question roster remains unidentified, so this measurement remains diagnostic.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3 Deep Think | 50.5 |