Anthropic has released Claude Sonnet 4.6, describing it as the most capable Sonnet model to date, with major upgrades across coding, long context reasoning, agent planning, computer use, and design.
The model introduces a 1 million token context window in beta, enabling users to process entire codebases, lengthy contracts, or large research archives in a single prompt.
Pricing remains unchanged from Sonnet 4.5 at $3 input and $15 output per million tokens, and Sonnet 4.6 is now the default model on Free and Pro tiers. Anthropic says performance that previously required an Opus class model is now available in the Sonnet tier.
Early users preferred Sonnet 4.6 over Sonnet 4.5 roughly 70% of the time and even favored it over Claude Opus 4.5 in 59% of tests, citing better instruction following, fewer hallucinations, and reduced overengineering.
The model also shows material gains in computer use. On the OSWorld benchmark, which tests real software interaction without APIs, Sonnet 4.6 approaches human-level performance on tasks such as navigating spreadsheets and completing multi-step web workflows. Anthropic said the release includes improved resistance to prompt injection attacks.
The launch follows Anthropic’s early February unveiling of its most advanced model, Claude Opus 4.6, positioning Sonnet 4.6 as a lower-cost alternative that narrows the capability gap.
