Claude Code: Claude Sonnet 4.6 Support

Claude Sonnet 4.6 Arrives in Claude Code

Claude Code 2.1.45, released on February 17, 2026, ships with first-class support for Claude Sonnet 4.6 — Anthropic's most capable Sonnet model to date. The release represents a comprehensive upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design.

Coding Performance That Rivals Opus

The most striking aspect of Sonnet 4.6 is how close it gets to Opus-level performance at a fraction of the cost. In Claude Code testing, users preferred Sonnet 4.6 over its predecessor Sonnet 4.5 roughly 70% of the time, citing better context reading and code consolidation. Even more notably, users preferred Sonnet 4.6 to Opus 4.5 — the frontier model from November 2025 — 59% of the time, rating it "significantly less prone to overengineering" and superior at instruction following.

This positions Sonnet 4.6 as the practical default for most development workflows in Claude Code. Anthropic recommends Opus 4.6 for tasks requiring the deepest reasoning, such as large-scale codebase refactoring and multi-agent coordination, but for day-to-day coding tasks Sonnet 4.6 delivers comparable results at unchanged pricing of $3/$15 per million tokens.

1 Million Token Context Window

Sonnet 4.6 introduces a 1M token context window, currently in beta, which doubles the previous Sonnet limit. For developers using Claude Code on large codebases — monorepos, complex multi-file refactors, or projects with extensive dependency chains — this expanded context means the model can hold significantly more of the codebase in working memory during a single session.

Context Compaction

A new beta feature called context compaction automatically summarizes older conversation context as discussions approach token limits. Rather than hitting a hard wall when the context fills up, the model intelligently compresses earlier parts of the conversation while preserving the most relevant details. This effectively extends the usable conversation length and maintains coherence across long coding sessions without requiring manual context management.

Computer Use and Agentic Capabilities

Beyond coding, Sonnet 4.6 shows major improvements in computer use and agentic workflows. The model achieves human-level capability on tasks like navigating complex spreadsheets and filling out multi-step web forms across multiple browser tabs. On the OSWorld benchmark, Sonnet 4.6 demonstrates significant gains over its predecessor. An insurance industry benchmark showed 94% accuracy on submission intake workflows — a result described as mission-critical for production deployments.

Safety

Anthropic's evaluations show Sonnet 4.6 is as safe as or safer than recent Claude models, with improved resistance to prompt injection attacks compared to Sonnet 4.5.