Scientific Reasoning on HLE (Score)
33.3ScoreGemini-3-pro High
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-3-pro Hightool=no tool, format=text only2026.06 | 33.3 | |
| Seed2.0 Protool=no tool, format=text only2026.06 | 32.4 | |
| GPT-5.2 Hightool=no tool, format=text only2026.06 | 29.9 | |
| Claude-Opus-4.5tool=no tool, format=text only2026.06 | 23.7 | |
| EAPO2025.09 | 16.07 | |
| Self-Exploratory RL2025.09 | 15.96 | |
| Claude-Sonnet-4.5tool=no tool, format=text only2026.06 | 14.5 | |
| Base Model2025.09 | 7 |