Anthropic releases Claude Sonnet 5, the most agentic model in the Sonnet line

On June 30, 2026, Anthropic released Claude Sonnet 5, the most agentic model in the Sonnet line. It approaches the quality of the flagship Claude Opus 4.8 but costs noticeably less.

What the new model can do

Claude Sonnet 5 can build plans, use tools such as a browser and a terminal, and work autonomously at a level that required larger, more expensive models just a few months ago. Sonnet-tier models (Claude Sonnet 3.5, 3.6, and 3.7) were the first to post impressive results in coding and tool use, and they effectively kicked off the era of agentic AI for developers. More recently, most of the gains in agentic capability have come from Opus-tier models, and Sonnet 5 is meant to close that gap.

• Release date: June 30, 2026, available on all Claude plans
• Predecessor: Claude Sonnet 4.6; quality benchmark: Claude Opus 4.8
• Launch API pricing: $2 per million input tokens and $10 per million output tokens until August 31, 2026, then $3 and $15 respectively
• Default model on the Free and Pro plans, available in Claude Code and on Claude Platform as claude-sonnet-5
• Tested on the BrowseComp agentic search benchmark and the OSWorld-Verified computer-use benchmark

How much safer is Sonnet 5 than the previous model?

Anthropic's evaluations show Sonnet 5 has a lower rate of undesirable behavior than Sonnet 4.6 and is safer overall in agentic use cases. Its ability to carry out cybersecurity tasks, however, is noticeably lower than that of the current Opus models.

How does Sonnet 5 compare with Opus 4.8?

Anthropic's charts show Sonnet 5 strictly improving on Sonnet 4.6 across effort levels. Opus 4.8 remains the model of choice where maximum accuracy matters, but Sonnet 5 gives developers noticeably better results at a lower price than was previously available. Users can adjust the effort level themselves to balance cost against accuracy, both within Sonnet 5 and when choosing between Sonnet 5 and Opus 4.8.

What it means

Anthropic is systematically pushing the agentic capabilities of its flagship models down to a more affordable tier. Six months ago, running an autonomous agent required the expensive Opus; now Sonnet is increasingly enough. That lowers the barrier to entry for developers building products on Claude.

FAQ

How much does Claude Sonnet 5 cost through the API?

Until August 31, 2026, the price is $2 per million input tokens and $10 per million output tokens. After that date, it rises to $3 and $15 respectively.

How does Sonnet 5 differ from Opus 4.8?

Opus 4.8 remains more accurate on complex agentic tasks, but Sonnet 5 delivers similar quality at a significantly lower price. The effort level is adjustable, letting you balance cost against accuracy.

Where is Claude Sonnet 5 available?

It is the default model on Claude's Free and Pro plans and is available to Max, Team, and Enterprise users, as well as in Claude Code and on Claude Platform via the API as claude-sonnet-5.