Anthropic is preparing for a notable week with indications that Claude Opus 4.5 could launch imminently, with some signs pointing to Monday as the potential release day. This timing aligns with recent benchmark leaks referencing the “Claude kayak” model, suggesting a possible internal or pre-release version was already tested.
The update is expected to target enterprise and advanced users who rely on top-tier reasoning and code understanding, as Claude Opus is typically positioned at the higher end of Anthropic’s lineup.
Parallel to this, Anthropic is expanding its feature set for Claude Skills. A new capability in development will allow users to request skill generation directly within the chat, removing the need for manual setup. Once generated, these skills can be enabled immediately or exported, streamlining the customization process and appealing to users looking for rapid automation within the Claude ecosystem.
New Claude web app changes
— Tibor Blaho (@btibor91) November 22, 2025
- new notice for "compaction status" ("Compacting our conversation so we can keep chatting. This takes about 1-2 minutes.")
- new "crimson lighthouse marble nux" banner announcement for the model selector dropdown with a new experiment "blue velvet… pic.twitter.com/cVSWHnDqjQ
A referral program for Claude Code is also surfacing, where users may receive three invite codes in the form of QR links to share. The requirements for participation remain undisclosed, but the approach points to a selective or early-access model designed to drive word-of-mouth adoption for Claude Code features.
Among upcoming updates, references to a “Megabrain” feature have been found, but its purpose is not yet clear. It could be related to model enhancements or entirely new capabilities, though specifics are speculative at this point. Overall, Anthropic appears to be finalizing several product announcements for Claude, positioning this week as potentially pivotal for both product launches and ecosystem growth.
