Claude Sonnet 4.6 Drops!
Claude Sonnet 4.6 delivers flagship-level computer use at mid-tier pricing. Outperforms on office tasks, coding, and agent planning. Now available for free.
“AI Disruption” Publication 8800 Subscriptions 20% Discount Offer Link.
Chinese New Year is truly the real battlefield for large models—the kind where the whole world joins in.
On the second day of the Lunar New Year, Anthropic released the strongest Sonnet in history—Claude Sonnet 4.6.
No more words—straight to the video.
It’s clear that computer use is the main highlight of this update.
Anthropic states that for tasks like filling out complex Excel sheets, web forms, and similar work, Sonnet 4.6 is already approaching human-level performance.
Other areas have seen all-around upgrades too: coding, long-context reasoning, agent planning, knowledge work, design... During the beta phase, it even supports a 1M token context window.
The key point: pricing remains exactly the same as Sonnet 4.5, and free users can use it too.
The cost-performance ratio is absurdly high.
On most agent tasks, Sonnet 4.6 performs roughly as well as the Opus series, but it’s faster and costs only 1/5 as much.
Anthropic says that internal testers’ preference for Sonnet 4.6 has already surpassed the giant-cup Opus 4.5.



