Anthropic Releases Claude Sonnet 4.6 With 1M-Token Context as New Default Model

Anthropic released Claude Sonnet 4.6 on February 17, 2026, immediately designating it the default model for Free and Pro plan users on claude.ai and Claude Cowork. The model delivers Opus-class performance across coding, computer use, long-context reasoning, agentic planning, and knowledge work—at the same price as its predecessor.

Key Features

The standout feature is a 1-million-token context window in beta, allowing developers to process entire codebases or dozens of research papers in a single prompt. The model also supports extended thinking mode. On the Pace insurance automation benchmark, it achieved 94% accuracy on computer use tasks, and matched Opus 4.6 on the OfficeQA enterprise document benchmark.

Performance

In early access evaluations, developers preferred Sonnet 4.6 over Sonnet 4.5 roughly 70% of the time. More notably, they preferred it over the larger Opus 4.5 model about 59% of the time—meaning tasks that previously required an Opus-class model can now be handled at Sonnet pricing.

Pricing

Pricing remains unchanged from Sonnet 4.5: $3 input / $15 output per million tokens. Safety evaluations confirmed it is as safe as, or safer than, other recent Claude models.

Source: Anthropic Official Announcement