Search on Humanity's Last Exam (HLE) (test)
45.8AccuracyGemini-3-pro
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-3-pro# Samples=?, Training=?, Academic=No2026.05 | 45.8 | |
| GLM-4.7-357B# Samples=?, Training=?, Academic=No2026.05 | 42.8 | |
| GPT-5-High# Samples=?, Training=?, Academic=No2026.05 | 41.7 | |
| DeepSeek-V3.2-671B# Samples=?, Training=?, Academic=No2026.05 | 40.8 | |
| OpenSeeker-v2-30B-SFT# Samples=10.6 k, Training=SFT, Academic=Yes2026.05 | 34.6 | |
| RedSearcher-30B# Samples=?, Training=CPT + SFT + RL, Academic=No2026.05 | 34.3 | |
| Tongyi DeepResearch# Samples=?, Training=CPT + SFT + RL, Academic=No2026.05 | 32.9 | |
| Claude-4.5-Sonnet# Samples=?, Training=?, Academic=No2026.05 | 32 | |
| WebSailor-V2-30B-RL# Samples=?, Training=SFT + RL, Academic=No2026.05 | 30.6 | |
| GLM-4.6-357B# Samples=?, Training=?, Academic=No2026.05 | 30.4 | |
| DeepSeek-V3.1-671B# Samples=?, Training=?, Academic=No2026.05 | 29.8 | |
| OpenAI Deep Research# Samples=?, Training=?, Academic=No2026.05 | 26.6 | |
| WebSailor-V2-30B-SFT# Samples=?, Training=SFT, Academic=No2026.05 | 23.9 | |
| OpenAI-o3# Samples=?, Training=?, Academic=No2026.05 | 20.2 |