Anthropic’s Claude Sonnet 4.6 Lands With a Massive Context Window Upgrade

Anthropic just pushed out a new version of its popular midsized AI model, and this one comes with some genuinely exciting upgrades. Claude Sonnet 4.6 dropped on February 17, 2026, right on schedule with the company’s four-month release cadence.

So what’s new? Anthropic focused this update on three areas: coding performance, instruction-following, and computer use capabilities. Plus, the model is now the default for both Free and Pro plan users, which means millions of people get these improvements automatically.

That Context Window Is a Big Deal

The headline feature here is storage capacity for information. Sonnet 4.6 ships with a beta context window of 1 million tokens. That’s twice the size of what the previous Sonnet offered.

Think of a context window like a whiteboard. The bigger it is, the more information the model can hold in front of it while working. At 1 million tokens, Anthropic says you can now fit entire codebases, lengthy contracts, or dozens of research papers into a single conversation.

For developers especially, that’s a meaningful upgrade. Instead of feeding code to the model in chunks, you can drop in a whole project and let it reason across the full picture. That changes how you work with it entirely.

Benchmark Scores Worth Noting

Sonnet 4.6 launched alongside a fresh set of benchmark results, and a few stand out. It set new records on OS World, which tests computer use, and on SWE-Bench, which measures software engineering performance.

But the most attention-grabbing number is its 60.4% score on ARC-AGI-2. This benchmark specifically tries to measure skills associated with human-level general intelligence, making it a tougher target than most standard tests. That score puts Sonnet 4.6 ahead of most comparable midsized models currently available.

However, it’s worth being straight about the competitive picture. Sonnet 4.6 still trails Anthropic’s own Opus 4.6, Google’s Gemini 3 Deep Think, and at least one refined version of GPT 5.2 on that same benchmark. So it’s strong, but it’s not leading the whole field.

Where This Fits in Anthropic’s Lineup

This release came just two weeks after Anthropic launched Opus 4.6, its more powerful flagship model. An updated Haiku model is expected to follow in the coming weeks, which would round out the full Claude lineup refresh.

Sonnet has always been Anthropic’s sweet spot model. It balances capability with accessibility, which is why it serves as the default for everyday users. This update reinforces that positioning, giving both casual users and developers something genuinely useful without requiring a premium tier upgrade.

If you’re already using Claude through the Free or Pro plan, you’re already running Sonnet 4.6. The expanded context window in beta is available now. Worth exploring if you’ve ever hit the limits of what you could load into a single session before.