Anthropic has released Claude Sonnet 5, an update that is substantially improved over predecessor Claude Sonnet 4.6 in coding, reasoning, tool use, and knowledge work, according to the company.
Claude Sonnet 5 was introduced June 30. It can make plans, use tools like browsers and terminals, and run autonomously at a level that required larger and more expensive models a few months ago, Anthropic said. Safety assessments found that Sonnet 5 shows an overall lower rate of undesirable behaviors than Sonnet 4.6 and is generally safer to use in agentic contexts, the company also noted.
Claude Sonnet 5 is a drop-in upgrade for Claude Sonnet 4.6 with three behavior changes: adaptive thinking is on by default, manual extended thinking now returns a 400 error (it was deprecated on Claude Sonnet 4.6), and setting sampling parameters to non-default values also returns a 400 error. Further, Sonnet 5 uses an updated tokenizer that changes how the model processes text to improve performance. Anthropic said Sonnet 5 offers performance close to that of Opus 4.8 model but at lower prices.
