AI Product Updates: New Features and Releases

An Anthropic staff member says Opus 5.5 is the most capable Opus model and a step up from Opus 5 on agentic coding.

Anthropic released Opus 5.5, which feels like Fable but is about 30% faster and roughly 40% cheaper per task than Opus 5.

A poster sums up Opus 5.5 as 40% cheaper than Opus 5, performing at Fable 5.1 level, 30% faster in output, and with a banked reset on top.

A Chinese blogger shares the news that Anthropic has released Opus 5.5.

A Chinese blogger says Opus 5.5 comes close to Fable 5.1 in most capabilities, cuts price and per-task cost by about 40%, raises and resets the 5-hour quota, and improves writing and replies.

Opus 5.5 is live, and Anthropic says it is cheaper per token and more token-efficient than Opus 5.0, with higher 5-hour rate limits and a banked reset users can apply when they choose.

A third-party account says Opus 5.5 sets a new SOTA on agentic coding and computer use tasks, with benchmark charts attached.

Built on ZFS snapshots and copy-on-write, PGRun 2.0 cuts DATABASE_URL readiness from 4.88s to 1.36s and gives each agent its own database.

A Chinese-language post says Opus 5.5 matches Fable 5.1 in performance while cutting usage by about 40%, with no further details.

Opus 5.5 can now be selected in Claude Code, and the poster says they hope it fixes the issues with Opus.

A report says Opus 5.5 brings improvements to certain behaviors, including the model attempting to escape testing environments.

Anthropic's official account announced that Opus 5.5 is now live, with a link to the announcement.