Anthropic Launches Claude Sonnet 5

Anthropic Claude Sonnet 5 approaches the performance of Opus 4.8 but costs much less to operate. Plus, it’s optimized for agentic use cases.

“Claude Sonnet 5 is built to be the most agentic Sonnet model yet,” Anthropic writes in the announcement post. “It can make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required larger and more expensive models.”

Sonnet is a cloud-hosted LLM, like Opus, but it’s faster, less expensive, and a better fit for most day-to-day AI interactions, Anthropic says. Opus, meanwhile, is better for more complex, multi-step reasoning activities that require very large context windows or teams of agents running in parallel.

With Sonnet 5, however, Anthropic is closing the functionality gap somewhat. It dramatically outscores its predecessor, Sonnet 4.6, in most of Anthropic’s internal tests while getting close to Opus 4.8 in agentic coding and computer use, and even outscoring it in knowledge work. Anthropic also notes that Sonnet 5 “shows an overall lower rate of undesirable behaviors than Sonnet 4.6, and is generally safer to use in agentic contexts.”

Claude Sonnet 5 is available today with introductory pricing of $2 per million input tokens and $10 per million output tokens. That will last through August 31, 2026, when the pricing shifts to $3 per million input tokens and $15 per million output tokens. Anthropic has also increased the rate limits across Chat, Cowork, Claude Code, and the Claude Platform to accommodate the higher token usage at higher effort levels, but users can select whichever level makes sense for them on a project-by-project basis.