Claude Sonnet 5, Anthropic's New Agentic Model

Anthropic launches Claude Sonnet 5, an agentic AI model outperforming Sonnet 4.6 in coding and reasoning at a lower cost for Free, Pro, and API users.

Anthropic is rounding out its Claude lineup with Sonnet 5, a model geared toward agentic uses. According to the publisher, it can plan, manipulate browsers and terminals, and operate autonomously at a level that recently required heavier and more expensive models. The model is touted as an improvement over its predecessor, Sonnet 4.6, in reasoning, tool use, coding, and knowledge work, with performance that the publisher likens to Opus 4.8, but at a lower cost.

Safety tests conducted prior to deployment indicate a lower rate of undesirable behaviors than Sonnet 4.6, as well as better resistance to prompt injection attempts. It comes with cyber safeguards activated by default, as its capabilities in this area are deemed by the publisher to be significantly inferior to those of Opus 4.8.

Sonnet 5 becomes the default model for Free and Pro offerings, and remains accessible to Max, Team, and Enterprise subscribers, both in Claude Code and via the API under the reference claude-sonnet-5. The platform is offering an introductory price of $2 per million input tokens and $10 per million output tokens until August 31, 2026, before increasing to $3 and $15.