Anthropic PBC introduced Claude Sonnet 5.5, a mid-tier model the company describes as its most capable to date for everyday work and a direct upgrade over the previous generation. Anthropic said the new model runs more than 30 percent faster than its predecessor while costing less to operate. The release follows the Sept. 22 launch of Opus 5.5, the flagship of the Claude family.
Anthropic designed Sonnet 5.5 for tasks including document generation, summarization, spreadsheets and debugging code. The company said the model also has a strong eye for design, can add polish to user interfaces and can follow templates to produce slides that need little editing. According to Anthropic, the model writes more clearly and in a more human-like style than earlier versions, and early testers called it faster to iterate on less complex tasks and smarter than the previous generation.
The model ships with invisible text watermarking. Anthropic said the watermarking does not affect text quality or readability but raises the likelihood that AI-generated text can be detected. The company added the capability to comply with global regulations, including the European Union's AI Act.
Pricing matches Sonnet 5 at $2 per million input tokens, $10 per million output tokens and 20 cents per million cache reads. Anthropic said the model typically needs far fewer tokens for the same work, making it 30 percent cheaper than the previous generation. Zendesk Inc. Director of AI Abhinay Kathuria said testing on hundreds of real support cases produced fewer wrong decisions and faster ticket resolution, with tickets processed 20 percent faster than the Claude models Zendesk runs in production.
On benchmarks, Anthropic said Sonnet 5.5 scored 70.6 percent on Terminal-Bench 4.0, an agentic coding evaluation, up from 10.3 percent for the previous model. It landed two points below Opus 5.5 on GDPval-AA, which tests real-world occupational work. Anthropic said it is the first Sonnet model to beat Pokémon Red using only screenshots, a game used to test long-horizon work and image understanding.
Anthropic said Sonnet 5.5 is the fastest Sonnet yet, with output speed more than 30 percent above the prior generation. Because it shares a foundation with Opus 5.5, it is the first Sonnet with notable cybersecurity capabilities and the first to launch with cyber safeguards and fallbacks like those built for Mythos-class models and Fable. Those safeguards may trigger on prompts about cybersecurity, biology or other sensitive content, though Anthropic said most software development and life sciences work will be unaffected. The company plans to release Haiku 5.5 in the coming weeks.
More AI news from TechManNews.
