Claude Sonnet 5 makes running AI agents a lot cheaper

Anthropic released Claude Sonnet 5 on June 30, 2026, and pitched it squarely at the thing that has been holding back real AI agents: cost. The company calls it the most agentic Sonnet it has shipped, able to make plans and use tools like browsers and terminals, and to run autonomously at a level that until recently required larger, pricier models. In short, you get performance close to Opus 4.8 for meaningfully less money.

That matters because agents are expensive to run. An agent that browses, calls tools, and iterates burns far more tokens than a single chat reply, so the per-token price decides whether an agentic feature is viable in production or just a demo.

Claude Sonnet 5 pricing

There is one asterisk worth knowing. Sonnet 5 ships with a new tokenizer that can turn the same text into up to about 35% more billable tokens, so the real-world cost isn't quite as low as the headline per-token figures suggest. Anyone budgeting seriously should test on their own workloads rather than assume a flat discount.

Sonnet 5 is now the default model on Anthropic's Free and Pro plans and is available to Max, Team, and Enterprise users, through Claude Code, and on the Claude Platform. For the many AI tools built on Anthropic's models, a cheaper, more capable default is the kind of quiet upgrade that makes agentic features affordable for everyone downstream.

Sources: Anthropic, TechCrunch, and Simon Willison's coverage, June 30, 2026.