Version 5.6.0

12 Apr 2026


Version 5.6.0
- Upgraded the premium AI model to GLM 5.1, Plus and Patron users should notice improved quality writing.
- Improved selection of compute provider based on their latency, failure rate, and cost.
- Reasoning efforts adjusted depending on use case (this controls how long the AI spend thinking before giving a response). It’s now set to medium for most chats, and high for rewrites, regenerations, outline openings, and for the summarisation system.
- More adjustments to the ban system, will be applying moderation to the backlog of existing chats shortly.
- Several minor security improvements, including tightening rate limiting and XSS hardening.
- Several bug fixes and optimisations, including fixing stale comet counts, reducing the number of database calls, fixing duplicate notifications, removing need for authenticated users to use cookies.
- Added ‘star’ and tag system to help sort private chats.
- Improvements to the summarisation system used to compress long AI chats.
- Age verification system tightened for jurisdictions that require it. (Australian and UK users now need to verify their age by entering payment information into Stripe before seeing any content marked as explicit, or using the AI).