Claude Sonnet 4.6

published

Anthropic's balanced model

Mid-tier model balancing capability and cost. Strong on coding and agentic tasks with 200k context.

Provider anthropic Type llm Access api Params undisclosed Context 200k License proprietary

Benchmarks (2)

Benchmark	Score	Reported By	Source
SWE-bench Verified	79.6%	vendor	source ↗
OSWorld	72.5%	vendor	source ↗

Best cost-performance ratio for production coding workloads. 72.5% on OSWorld demonstrates strong agentic capability.

Smaller context than Opus. No extended thinking mode. Proprietary.

Provider anthropic

Released 2026-02-17

Training cutoff 2025-04

Created March 22, 2026

Last reconciled Never