Claude Sonnet 4.5: benchmarks and ranking
Claude Sonnet 4.5 (Anthropic) — overall rank: #98 / 153, overall score: 33.7 (90% interval 27.6–40.5), leaderboards: 5. Released: 2025-09-29 · Price (in/out): $3 / $15
Rank on each leaderboard
| Leaderboard | Rank | Published value |
|---|---|---|
| Artificial Analysis | #80 / 141 | 20.7 |
| LiveBench | Not rated | — |
| Epoch AI | #79 / 230 | 147.56 |
| ARC Prize (ARC-AGI) | #59 / 95 | 13.6 |
| Scale AI SEAL | #24 / 44 | 13.72 |
| Vals AI | Not rated | — |
| Terminal-Bench | Not rated | — |
| EQ-Bench | Not rated | — |
| SimpleBench | #59 / 112 | 54.3 |
| BenchLM.ai | Not rated | — |
Domain scores
- Coding: 44.3
- Agents: 36.9
- Reasoning: 40.4
- Math: 49.0
- Knowledge: 62.3
- Writing: 63.8
- Instructions: 37.1
- Chat & EQ: 33.1
- Business: 70.1
- Japanese: 70.6
- Vision: 28.1
- Long context: 23.9