Anthropic on Tuesday introduced Claude Sonnet 5, describing it as its “most agentic Sonnet yet” and claiming it can make plans, use tools such as browsers and terminals, and run autonomously at a level that “just a few months ago required larger and more expensive models.”
The company also claimed Sonnet 5 is “a substantial improvement” over Sonnet 4.6 in reasoning, tool use, coding, and knowledge work, while landing “close to Opus 4.8” at lower prices. In a benchmark card shared alongside the launch, Sonnet 5 posted 63.2% on SWE-bench Pro versus 58.1% for Sonnet 4.6 and 69.2% for Opus 4.8; 80.4% on Terminal-Bench 2.1 versus 67.0% and 82.7%; and 81.2% on OSWorld-Verified versus 78.5% and 83.4%. On Humanity’s Last Exam with tools, Sonnet 5 scored 57.4%, just behind Opus 4.8’s 57.9%, while on GDPval-AA v2 it reached 1618, slightly ahead of Opus 4.8’s 1615.
Anthropic says early access partners found that Sonnet 5 finishes complex tasks where earlier Sonnets stopped short, checks its own output without being asked, and does its agentic work at what the company calls an “attractive price point.” The model is now the default on Free and Pro, and is also available to Max, Team, and Enterprise users across Claude apps and the Claude Platform.
The pricing table attached to the announcement lists Sonnet 5 at $2 per million input tokens and $10 per million output tokens through August 31, 2026, before rising to $3 and $15 on September 1, 2026, the same price as Sonnet 4.6. Anthropic also lists Sonnet 4.6 at the same standard $3/$15 rates and Opus 4.8 at $5/$25. Responses on X quickly turned to comparisons with Opus, pricing questions, and requests for higher usage limits.
Source: Claude (@claudeai)
