Source : INDIA TODAY NEWS
Anthropic is on a roll to bring out some of the most advanced AI models right now. Weeks after the debut of Claude Opus 4.6, the company has now launched a new AI model called Claude Sonnet 4.6. According to the company, this is its most capable Sonnet model so far, especially when it comes to coding and reasoning. Sonnet 4.6 will now also become the default model inside Anthropic’s Claude chatbot for both free users and Pro subscribers.
advertisement
Right now, Anthropic is making Sonnet 4.6 available across its chatbot platform. Free users get limited access that resets every five hours, while Pro users continue on the existing subscription pricing. Beyond the chatbot, Anthropic is also offering the Sonnet 4.6 API and on major cloud platforms, making it accessible to developers and enterprises building AI-powered tools.
“It’s a full upgrade of the model’s skills across coding, computer use, long-context reasoning, agent planning, knowledge work, and design. Sonnet 4.6 also features a 1M token context window in beta,” says the company in its official announcement blog post.
Capabilities of Claude Sonnet 4.6
So what’s new here? According to Anthropic, Sonnet 4.6 delivers stronger performance in coding, reasoning, and what it calls “knowledge work”, tasks such as analysing documents, working with spreadsheets, summarising reports, or assisting with design workflows. The company says the model is more consistent at following coding instructions and better at generating usable code without veering off track, an area where earlier AI systems often struggled.
The company claims that the Sonnet 4.6 model also improves reliability when writing, editing, and debugging code, and early testers reportedly preferred it over previous versions. In some internal tests, the model even reportedly outperformed Claude Opus 4.6 on specific agentic tasks.
In addition, Sonnet 4.6 is said to handle large volumes of data more effectively and process long documents due to its 1 million token context window, which is currently available in beta. Anthorpic notes that this expanded context allows the model to “remember” and analyse significantly more information in a single session, which could be particularly useful for legal documents, financial records, or large codebases.
Sonnet 4.6 excels in benchmarks
On benchmarks, Anthropic says Sonnet 4.6 performed strongly on evaluations such as Humanity’s Last Exam, GPQA Diamond, and SWE-bench Verified, tests commonly used to measure reasoning depth and coding accuracy. The company also pointed to results in insurance and enterprise automation use cases, where it reportedly saw improved performance over earlier Claude models.
As for safety, Anthropic adds that improvements remain part of the upgrade. The company says Sonnet 4.6 shows lower rates of hallucination and reduced “sycophancy”, the tendency of AI systems to agree too easily with user assumptions even when they are incorrect.
Interestingly, Anthropic’s rapid push in AI comes at a time when the race is accelerating. Companies like OpenAI and Google are also rapidly iterating on their flagship models, each pushing improvements in reasoning, coding, and multimodal capabilities. Amid this intense race, companies are now rolling out major model updates at a much faster pace, with new versions arriving weeks apart rather than months.
– Ends
SOURCE :- TIMES OF INDIA
