Claude Sonnet 4.5
Claude Sonnet 4.5 is an API-only (closed weights) language model from Anthropic, released September 29, 2025. The context window is 195K tokens. On the Epoch Capabilities Index it scores 147.6, rank 80 of 256 rated models. API price: $3 input / $15 output per million tokens.
Claude Sonnet 4.5 is the best coding model in the world. It's the strongest model for building complex agents. It’s the best model at using computers. And it shows substantial gains in reasoning and math. Code is everywhere. It runs every application, spreadsheet, and software tool you use.
Check it against your own PC →
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Reasoning | 12.0 | Chess Puzzles, SimpleBench, Mystery Game Puzzles, Humanity's Last Exam |
| Math | 32.1 | ProofBench, MATH level 5, OTIS Mock AIME 2024-2025, FrontierMath-Tiers-1-3-v2-Private, FrontierMath-Tier-4-v2-Private |
| Coding | 44.7 | GSO-Bench, Terminal Bench, SWE-bench Verified |
| Agent | 31.2 | OSWorld, Remote Labor Index |
| Knowledge | 65.5 | GPQA diamond, SimpleQA Verified |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MATH level 5 | 97.7% |
| GPQA Diamond | 79.6% |
| OTIS Mock AIME | 77.8% |
| SWE-bench Verified | 71.3% |
| OSWorld | 62.9% |
| Terminal-Bench | 46.5% |
| SimpleBench | 45.2% |
| SimpleQA Verified | 30.7% |
| FrontierMath (Tiers 1–3) | 25.1% |
| ProofBench | 19.0% |
| GSO-Bench | 14.7% |
| Humanity's Last Exam | 13.7% |
| Mystery Game Puzzles | 8.6% |
| Chess Puzzles | 5.3% |
| FrontierMath-Tier-4-v2-Private | 2.4% |
| EBR-bench | 2.4% |
| Remote Labor Index | 2.1% |
Links
Announcement · Epoch AI benchmarks
Models with a similar score
Kimi K2.6 · GPT-5 Pro · Kimi K2.7 Code · GPT-5.1 · Beam 1.0 · Inkling-Small · GLM-5.1 · DeepSeek-V4-Pro
More from Anthropic
Claude Fable 5.2 · Claude Haiku 5.5 · Claude Sonnet 5.5 · Claude Opus 5.5 · Claude Fable 5.1 · Claude Opus 5 · Claude Sonnet 5 · Claude Fable 5