Agents' Last Exam Unlicensed Overall Pass Rate
Agents · 2026-06-03
Perfect-score pass rate across the 141 Agents' Last Exam tasks that do not require licensed software, paid APIs, or proprietary data.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Sol | 33.0 |
| GPT-5.6 Luna | 31.9 |
| GPT-5.6 Terra | 29.8 |
| Kimi K3 | 29.8 |
| Opus 4.8 | 29.1 |
| GPT-5.5 | 28.7 |
| Claude Fable 5 | 27.0 |
| GPT-5.4 | 22.1 |
| Opus 4.7 | 22.0 |
| Composer 2.5 | 22.0 |
| GLM-5.2 | 22.0 |
| Doubao Seed 2.1 Pro | 19.9 |
| Gemini 3.1 Pro Preview | 17.0 |
| DeepSeek-V4-Pro | 13.4 |
| Qwen3.7-Max | 12.8 |
| GLM-5.1 | 12.4 |
| Kimi K2.6 | 9.9 |
| MiMo-V2.5 | 9.2 |
| Qwen3.6 Plus (2026-04-02) | 9.2 |
| Grok 4.3 | 7.1 |
| MiniMax M2.7 | 6.4 |