Claude Sonnet 5.5 Arrives in Claude Code as the New Default Sonnet
Claude CodeClaude Code 2.1.284 added Claude Sonnet 5.5 (claude-sonnet-5-5), which is now the default Sonnet model on the Anthropic API. The model keeps a 1M-token context window and is priced at $2 per million input tokens and $10 per million output tokens, with cache reads at $0.20 per million tokens. Anthropic describes it as a clear upgrade over Sonnet 5 that runs 30% faster and costs up to 30% less.
Key Takeaways
Claude Sonnet 5.5 (claude-sonnet-5-5) is now the default Sonnet model on the Anthropic API as of Claude Code 2.1.284.
Anthropic describes it as a clear upgrade over Sonnet 5 that runs 30% faster and costs up to 30% less.
The 1M-token context window carries over, so long sessions and large codebases stay in context.
Pricing is $2 per million input tokens and $10 per million output tokens, the same list rates as Sonnet 5.
Cache reads cost $0.20 per million tokens, which lowers the cost of long agentic sessions that reuse context.
The launch comes days after Claude Opus 5.5, refreshing both main model tiers in Claude Code in one week.
A New Default Sonnet
Claude Code 2.1.284, released on September 28, 2026, added support for Claude Sonnet 5.5, identified by the model ID claude-sonnet-5-5. On the Anthropic API it now serves as the default Sonnet model, so anyone who selects the Sonnet alias in Claude Code moves to the new version without changing a setting.
Anthropic announced the model the same day and positioned it as a clear upgrade over Sonnet 5, running 30% faster and costing up to 30% less. The announcement did not include a detailed benchmark table in the newsroom listing, so the concrete numbers below come from the Claude Code release notes.
Context Window and Pricing
Sonnet 5.5 keeps the 1M-token context window that Sonnet 5 introduced as a native feature of Claude Code. Long sessions, large repositories and heavy tool output can stay in context without a separate long-context tier.
Pricing is listed at $2 per million input tokens and $10 per million output tokens. Cache reads cost $0.20 per million tokens, which matters most in long agentic sessions where the same context is read again and again. The rates match the introductory pricing Sonnet 5 launched with, a level Anthropic later made permanent for Sonnet 5.
What It Means for Claude Code Users
For day-to-day work, the practical effect is a faster and cheaper default for the mid-tier model. Users on plans where Sonnet is the default get the newer model automatically. Users who pin a specific model through settings or environment variables keep what they pinned. Everyone else can switch with the /model picker or the sonnet alias.
The release lands days after Claude Opus 5.5, which became the new default Opus model on September 22. Together the two releases refresh both of the main model tiers in Claude Code within a single week.