GPT-5.6 Sol sweeps all four Agents' Last Exam splits
Sol ranks first across the four views among 20 models with scores, while Claude Fable 5 remains in the top five on each.
Event date Published
Sol leads every Agents' Last Exam split
Overall
GPT-5.6 Sol (#1) 53.62
Claude Fable 5 (#4) 48.7
Near-term
GPT-5.6 Sol (#1) 78.82
Claude Fable 5 (#5) 71.13
Full-Spectrum
GPT-5.6 Sol (#1) 46.58
Claude Fable 5 (#3) 42
Last-Exam
GPT-5.6 Sol (#1) 19.42
Claude Fable 5 (#2) 16.5
| Panel | Item | ALE score (%) |
|---|---|---|
| Overall | GPT-5.6 Sol (#1) | 53.62 |
| Overall | Claude Fable 5 (#4) | 48.7 |
| Near-term | GPT-5.6 Sol (#1) | 78.82 |
| Near-term | Claude Fable 5 (#5) | 71.13 |
| Full-Spectrum | GPT-5.6 Sol (#1) | 46.58 |
| Full-Spectrum | Claude Fable 5 (#3) | 42 |
| Last-Exam | GPT-5.6 Sol (#1) | 19.42 |
| Last-Exam | Claude Fable 5 (#2) | 16.5 |
Overall is the aggregate view; Near-term, Full-Spectrum, and Last-Exam are the source's published splits. Among 20 models with scores, Sol leads all four. Its selected settings are xhigh, max, xhigh, and medium in that order.