Anthropic Releases Claude Sonnet 5 as Default Model, Trails Its Own Opus on Agentic Coding Benchmark

Anthropic released Claude Sonnet 5, a more powerful and agentic version of its midsize model. TechCrunch reported that the model can make plans, use tools such as browsers and terminals, and run autonomously. Starting Tuesday, Claude Sonnet 5 became the default model for free and Pro plans and is available for every subscription.

At launch the model is priced at $2 per million input tokens and $10 per million output tokens through August 31, after which input pricing rises to $3 per million tokens while output pricing stays at $10 per million. On an agentic coding benchmark Sonnet 5 scored 63.2 percent, compared with 69.2 percent for Opus 4.8 and 58.1 percent for Sonnet 4.6.

The new model slightly outperformed Opus 4.8 on a knowledge work benchmark.

It also showed a lower rate of undesirable behaviors such as cooperation with misuse and deception than Sonnet 4.6, and a much lower ability to perform dangerous cybersecurity tasks than current Opus models. Daniel Shepard, senior engineer at Zapier, stated that Claude Sonnet 5 completed a two-part job updating Salesforce account tiers and sending a launch announcement to enterprise contacts end to end.

Fabian Hedin, co-founder of Lovable, stated that Claude Sonnet 5 refuses unsafe requests cleanly and consistently.

OpenAI launched GPT-5.6 Sol in preview last week, and Google launched Gemini 3.5 Flash in May. TechCrunch reported that Sonnet 5 is positioned as a cheaper alternative to Opus 4.8, GPT-5.5, and Gemini 3.1 Pro.