Claude Sonnet 5.5: benchmarks and ranking
Claude Sonnet 5.5 (Anthropic) — overall rank: #13 / 153, overall score: 86.7 (90% interval 84.1–88.9), leaderboards: 8. Released: 2026-09-28 · Price (in/out): $2 / $10
Rank on each leaderboard
| Leaderboard | Rank | Published value |
|---|---|---|
| Artificial Analysis | #2 / 141 | 56 |
| LiveBench | #21 / 64 | 77.8 |
| Epoch AI | #10 / 230 | 166.6 |
| ARC Prize (ARC-AGI) | #20 / 95 | 83.9 |
| Scale AI SEAL | Not rated | — |
| Vals AI | #2 / 42 | 70.9 |
| Terminal-Bench | #5 / 53 | 63.94 |
| EQ-Bench | Not rated | — |
| SimpleBench | #21 / 112 | 75.9 |
| BenchLM.ai | #13 / 24 | 77.7 |
Domain scores
- Coding: 92.3
- Agents: 90.3
- Reasoning: 84.6
- Math: 82.3
- Knowledge: 78.2
- Writing: 75.5
- Instructions: 68.1
- Chat & EQ: 53.2
- Business: 86.9
- Japanese: 81.7
- Vision: 68.4
- Long context: 58.1