Anthropic released Claude Sonnet 5 on 30 June 2026. It is now the default model for Free and Pro users, and available across Claude Code, the Claude API and Anthropic's platform under the name claude-sonnet-5. Anthropic is calling it the most agentic Sonnet model yet, meaning it plans multi-step tasks and uses tools like browsers and terminals with less hand-holding than the previous version.
The Numbers That Matter
On Humanity's Last Exam with tools enabled, Sonnet 5 scores 46.8%, up from Sonnet 4.6's 34.6%. On OSWorld-Verified, a benchmark for computer-use tasks, it hits 78.5%. Anthropic says it approaches Opus 4.8 level performance at a fraction of the cost. Pricing is $2 per million input tokens and $10 per million output tokens through 31 August 2026, rising to $3 and $15 after that.
Why the Introductory Price Is the Real Story
Benchmark jumps matter to developers. The pricing window matters to anyone running an AI feature on a budget. For two months, a model close to Anthropic's top tier is priced closer to its cheap tier. If you have been holding off building something because the good models cost too much per call, this is the window to build it and lock in usage patterns before the standard pricing kicks in on 1 September.
What To Do Before End of August
If your business runs anything on the Claude API, from a support chatbot to a document summariser, check which model it is calling. Sonnet 5 at introductory pricing is a genuine upgrade in capability at a lower cost than many businesses are paying for older models right now. Test it against your current setup before the price rises in September, not after.