Written By Shubham Arora
Published By: Shubham Arora | Published: Feb 18, 2026, 05:08 PM (IST)
AI company Anthropic has introduced Claude Sonnet 4.6, calling it the most capable Sonnet model so far. The update brings improvements across coding, reasoning, and computer-use tasks, and it is now available to both free and paid users. Also Read: Apple’s next devices could be AI smart glasses and AI pendant: What we know
The announcement was shared through Anthropic’s official blog, where the company said Sonnet 4.6 is designed to handle more advanced work while remaining practical for everyday use. Also Read: India AI Impact Summit 2026: Infosys teams up with Anthropic for enterprise AI agents
One of the key updates in Claude Sonnet 4.6 is its 1 million token context window, currently in beta. This allows the model to process and reason over much longer inputs, including large codebases, lengthy documents, or multiple files in a single session without resetting. Also Read: Inside India AI Impact Summit 2026: PM Modi to global CEOs - what happened on day 1
Anthropic says developers who tested the model early preferred it over the previous Sonnet 4.5 version. According to the company, users reported better instruction following, fewer hallucinations, and more consistent outputs. In internal testing cited by Anthropic, users chose Sonnet 4.6 over Sonnet 4.5 roughly 70 percent of the time.
Interestingly, the company also claimed that some users preferred Sonnet 4.6 even over its earlier Opus 4.5 model in certain coding and instruction-heavy tasks.
Anthropic has been pushing its “computer use” feature, where the AI interacts with software in a way similar to a human. Instead of relying only on APIs, the model can click, type, and navigate across interfaces.
As mentioned in the official release, Sonnet 4.6 shows improved performance on OSWorld, a benchmark that tests AI interaction across real software such as browsers and productivity tools. The company says users are seeing better results in tasks like navigating spreadsheets or completing multi-step web forms.
Claude Sonnet 4.6 is now the default model on claude.ai and Claude Cowork for Free and Pro users. Pricing remains the same as Sonnet 4.5, starting at $3 per million input tokens and $15 per million output tokens.
The model is also available via the Claude API and major cloud platforms. Developers can access it using the “claude-sonnet-4-6” model name.
Anthropic said it conducted safety evaluations before launch. According to the company’s blog, researchers found no major safety concerns and described the model as showing strong safety behaviour.