Claude Fable 5.2: benchmarks and ranking
Claude Fable 5.2 (Anthropic) — overall rank: #2 / 153, overall score: 92.2 (90% interval 90.3–93.8), leaderboards: 6. Released: 2026-11-09
Rank on each leaderboard
| Leaderboard | Rank | Published value |
|---|---|---|
| Artificial Analysis | Not rated | — |
| LiveBench | Not rated | — |
| Epoch AI | #1 / 230 | 173.4 |
| ARC Prize (ARC-AGI) | #5 / 95 | 93.8 |
| Scale AI SEAL | Not rated | — |
| Vals AI | #1 / 42 | 71.3 |
| Terminal-Bench | #2 / 53 | 65.76 |
| EQ-Bench | Not rated | — |
| SimpleBench | #1 / 112 | 94.9 |
| BenchLM.ai | #1 / 24 | 93.3 |
Domain scores
- Coding: 94.3
- Agents: 95.9
- Reasoning: 94.5
- Math: 91.2
- Knowledge: 96.7
- Writing: 95.9
- Instructions: 95.2
- Chat & EQ: 59.2
- Business: 81.7
- Vision: 94.8