Introducing Claude Sonnet 4.6. Sonnet 4.6 delivers frontier performance across coding, agents, and professional work at scale.

Anthropic announced the release of Claude Sonnet 4.6 on February 17, 2026, positioning the new model to deliver capabilities previously associated with its top-tier offerings at a more accessible price. The update is now the default model for users on claude.ai’s free and pro plans, aiming to make high-performance AI practical for a wider range of coding, office automation, and knowledge work tasks without requiring customers to pay for the company's most expensive models.

Sonnet 4.6 features significant technical upgrades, including a 1M token context window in beta and marked improvements in computer use, as measured by the OSWorld-Verified benchmark. In early testing for coding tasks, developers preferred Sonnet 4.6 over its predecessor, Sonnet 4.5, approximately 70% of the time and even favored it over the November 2025 Opus 4.5 model in 59% of cases. Pricing remains unchanged from the previous Sonnet version, starting at $3 per million input tokens, while safety evaluations indicate improved resistance to prompt injection attacks.

This release intensifies competition in the AI market by compressing the performance gap between mid-tier and frontier models. By pushing Opus-level document comprehension and agentic planning capabilities down to the more economical Sonnet line, Anthropic is directly addressing the cost-performance trade-off that often limits enterprise adoption of complex AI workflows. Customer evaluations from firms like Box, which reported a 15-point performance increase on reasoning tasks, suggest the model is effective for real-world business applications, potentially accelerating the deployment of AI agents in corporate environments.

Anthropic's release of Sonnet 4.6 signals an acceleration in the commoditization of near-frontier AI capabilities, shifting the competitive focus to the performance-per-dollar of mid-tier models for scalable enterprise deployment.