Anthropic Releases Upgraded Claude Sonnet 4.6

Anthropic released Claude Sonnet 4.6 on Tuesday, introducing significant upgrades to its mid-tier AI model that bring performance levels closer to the company's flagship Opus series while maintaining the same pricing structure.

The new version, described by Anthropic as the most capable Sonnet model to date, features a beta 1 million token context window. This expansion allows the AI to process vast amounts of information in a single prompt, such as entire codebases, lengthy contracts or dozens of research papers. Developers and users can now handle complex tasks requiring deep context without needing to fragment inputs, a capability previously limited to higher-end models.

Improvements span multiple areas, including coding, long-context reasoning, agent planning, computer use, knowledge work and design. Anthropic highlighted enhanced instruction following, reduced hallucinations and less overengineering compared to the prior Sonnet 4.5. In internal evaluations, early users preferred Sonnet 4.6 over Sonnet 4.5 in about 70 percent of tests and even favored it over the earlier Claude Opus 4.5 in 59 percent of cases. The model excels in bug detection, code fixes and architecture...