Agents' Last Exam Overall Score
Agents · 2026-06-03
Average partial-credit rubric score across the full 152-task Agents' Last Exam set, reported separately from the stricter perfect-score pass rate.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 63.2 |
| GPT-6 Astra | 59.3 |
| GPT-6 Sol | 58.6 |
| Claude Opus 5 | 55.9 |
| Muse Spark 1.3 | 55.8 |
| MiniMax M3.1 Flash | 55.0 |
| GPT-5.6 Sol | 53.6 |
| Kimi K3 | 51.6 |
| GPT-6 Luna | 50.9 |
| GPT-5.6 Terra | 50.7 |
| GPT-5.6 Luna | 50.4 |
| Claude Fable 5 | 48.7 |
| GPT-5.5 | 47.9 |
| Opus 4.8 | 45.2 |
| Agnes 2.5 Pro Beta | 42.7 |
| Opus 4.7 | 42.5 |
| Doubao Seed 2.1 Pro | 41.9 |
| GLM-5.2 | 41.1 |
| Composer 2.5 | 39.5 |
| GPT-5.4 | 37.3 |
| Gemini 3.1 Pro Preview | 32.7 |
| Qwen3.7-Max (source unspecified) | 31.6 |
| GLM-5.1 | 28.1 |
| DeepSeek-V4-Pro | 27.6 |
| Qwen3.6 Plus (2026-04-02) | 24.6 |
| MiMo-V2.5 | 23.6 |
| Kimi K2.6 | 21.7 |
| Grok 4.3 | 21.3 |
| MiniMax M2.7 | 14.2 |