Agentic Search on HLE text-only
45.8Pass@1Gemini 3-pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3-proevaluation_set=full set2026.03 | 45.8 | |
| GPT-5 High2026.03 | 41.7 | |
| Seed1.82026.03 | 40.9 | |
| Claude Sonnet-4.52026.03 | 32 | |
| Gemini 2.5-pro2026.03 | 19.8 |
Loading the SOTA2 catalog…