JAKARTA - Anthropic launched the latest version for Claude, namely Sonnet 4.6. This latest model is claimed to have superior capabilities in working on complex coding and reasoning.
Sonnet 4.6 is now the default model set in Claude.ai, for both free and paid users. Anthropic said that the subscription price would not change even though the platform was getting performance improvements.
This model brings many improvements, one of which is a larger context window. In the beta version, users can get a capacity of up to one million tokens. This capacity can process the entire code base in one request.
In internal testing conducted by Anthropic, Sonnet 4.6 proved more effective in reading the context before modifying the code. Developers also noted that this model rarely makes mistakes compared to the Claude Opus 4.5 model.
"Sonnet 4.6 has characters that are generally warm, honest, prosocial, and sometimes funny, very strong security behaviors, and no signs of major concerns around high-risk forms of nonconformity," the developers wrote in the test results, quoted on Wednesday, February 18.
The reasoning ability of this model is also tested through a business simulation called Vending-Bench Arena. The results show that Sonnet 4.6 is able to develop a smart long-term investment strategy to generate maximum profit.
For office users, Anthropic presents a new integration through an Excel add-in that supports the MCP connector. This feature allows Claude to retrieve data from external sources such as Moody's and FactSet without leaving the spreadsheet.
Claude Sonnet 4.6 is now available globally on various platforms starting today. Users can immediately experience Claude's performance improvement through the API and Claude mobile application, both on Android and iOS.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
