Multi-domain Knowledge and Reasoning on HLE (official)
42Exact MatchGPT-5 pro
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5 protools=with tools2026.02 | 42 | |
| ChatGPT Agent2026.02 | 41.6 | |
| Tendem’s AI agenthuman involvement=none2026.02 | 39 | |
| GPT-5 hightools=with tools2026.02 | 35.2 | |
| ChatGPT Deep Research2026.02 | 26.6 | |
| o3 hightools=with tools2026.02 | 24.3 | |
| Perplexity Deep Research2026.02 | 21.1 |