Anthropic Releases Claude Sonnet 5 With Enhanced Agentic Capabilities

Anthropic released Claude Sonnet 5 this week, positioning it as its most agentic Sonnet model to date. This new iteration is designed to plan, operate browsers and terminals, and execute tasks autonomously over extended periods. Sonnet 5 is now the default model for Anthropic's Free and Pro subscription plans and is also available to Max, Team, and Enterprise users, as well as within Claude Code and on the Claude Platform.

In performance comparisons, Sonnet 5 demonstrates significant improvements over its predecessor, Sonnet 4.6, which was released in February 2026. Anthropic reports that Sonnet 5 surpasses Sonnet 4.6 across all published benchmarks, achieving 63.2% on SWE-bench Pro, 81.2% on OSWorld-Verified, and 57.4% on HLE. The company emphasizes that the release focuses on agentic reliability, which translates to maintaining context over longer task chains and exhibiting better self-correction when tool calls fail.

Economically, Sonnet 5 presents a more cost-effective option compared to the flagship Opus 4.8 model. Introductory API pricing for Sonnet 5 is set at $2 per million tokens for input and $10 per million tokens for output through August 31, after which it will increase to $3 and $15, respectively. In contrast, Opus 4.8 is priced at $5 per million tokens for input and $25 per million tokens for output. Anthropic suggests Sonnet 5 offers the best value for low to medium effort tasks, though it may incur higher costs than Opus 4.8 for similar quality at extra-high effort levels.

Anthropic also highlights that Sonnet 5 is designed to be safer than Sonnet 4.6, with deliberately limited cyber capabilities. This distinction positions Opus 4.8 as the preferred choice for accuracy-critical applications. Sonnet 5 utilizes an updated tokenizer, the same one introduced with Opus 4.7, which can result in a 1.0 to 1.35 times increase in token count for the same text. Sonnet 5 sits between the more affordable Haiku 4.5 and the top-tier Opus 4.8 in Anthropic's model hierarchy.