Anthropic Advances AI Capabilities with Sonnet 4.6 Release

Anthropic has launched Sonnet 4.6, an updated version of its mid-size AI model, continuing its established four-month release cycle. This iteration introduces significant enhancements in coding proficiency, instruction-following, and computer use, aiming to improve user experience for both Free and Pro plan subscribers, for whom Sonnet 4.6 will become the default model.

A standout feature of Sonnet 4.6 is its expanded context window of 1 million tokens, doubling the previous maximum and enabling the model to process entire codebases, lengthy contracts, or multiple research papers within a single request. This capacity marks a notable advancement in handling extended and complex inputs.

The release follows closely on the heels of Anthropic's Opus 4.6 launch and precedes an expected update to the Haiku model. Benchmark tests reveal new performance records, including top scores on OS World (computer use) and SWE-Bench (software engineering). Most notably, Sonnet 4.6 achieved a 60.4% score on ARC-AGI-2, a benchmark designed to assess human-like intelligence skills. While this score surpasses many comparable models, it remains behind leading AI systems such as Opus 4.6, Gemini 3 Deep Think, and a refined GPT 5.2 variant.

These developments reflect Anthropic's ongoing commitment to advancing AI capabilities, particularly in domains requiring complex reasoning and extensive data processing.

Read more: techcrunch.com
Infosys Collaborates with Anthropic to Develop Enterprise-Grade AI Agents Amid Industry Shifts