Claude Sonnet 5: Opus-class agents at Sonnet prices

Anthropic released Claude Sonnet 5 on June 30, 2026. It's positioned as the most agentic Sonnet yet — planning, browser and terminal use, and autonomous runs that until recently required Opus-class models. Anthropic states performance is close to Opus 4.8 at lower prices, and calls Sonnet 5 a strict improvement over Sonnet 4.6 on the BrowseComp agentic search and OSWorld-Verified computer use evaluations. Introductory pricing through August 31, 2026 is $2 per million input tokens and $10 per million output tokens; after that, $3 and $15. Opus 4.8 sits at $5 and $25. Sonnet 5 is the default on Free and Pro, available on Max, Team, Enterprise, in Claude Code, and via the Claude API as claude-sonnet-5. Anthropic reports "an overall lower rate of undesirable behaviors than Sonnet 4.6" and a much lower ability to perform cybersecurity tasks than current Opus models. Early access testers describe end-to-end task completion where prior Sonnets stalled — one example: updating Salesforce account tiers and sending a launch announcement to enterprise contacts in a single run. Another tester reports the model wrote a reproducing test, implemented a fix, then stashed the change to confirm the bug returned — unprompted.