Claude Sonnet 5.5 is the everyday complement to Opus 5.5: strongest at well-scoped tasks such as fixing bugs and producing polished documents, slides and spreadsheets. It is a large step up from Sonnet 5 on agentic work (Terminal-Bench 4.0 70.6% vs 10.3%; OSWorld 2.1 80.1% vs 57.0%) and matches Opus 5.5 on knowledge-work evaluations, while staying at $2/$10 per million tokens. Adaptive thinking is on by default and can be limited to between tool calls. It accepts text and images with a 1M-token context window and up to 128K output tokens.
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| SciCode | 61.0 | accuracy | today |
| TerminalBench | 63.6 | accuracy | today |
| Humanity's Last Exam | 55.0 | accuracy | today |
| LCR | 82.7 | accuracy | today |
Updated today
Predecessors
Claude Sonnet 5