A cleaner audit log, tighter checks
Choose how deeply the model thinks in each Workspace chat, filter the audit log to Workspace activity, read small costs in full, pay list price for image generation, watch product films in the docs, and rely on images, cached answers and agency two-factor rules behaving as set. Owners and admins also accept a clearer Data Processing Agreement once. Everyone also accepts an updated Privacy Policy once, which now says which Sluis apps you use.
- New
Choose a reasoning level per chat
Next to the model in the composer you can now set how much the model thinks before it answers: Off, Low, Medium or High. Off is fastest, High is most thorough. Each chat remembers its choice. Leave it on Automatic and Sluis decides per question, as before. The control is greyed out for models that cannot reason.
- Fixed
Image generation billed at list price
OpenAI image models such as gpt-image-2 and gpt-image-2.5 were billed far above their list price. They now bill text and image tokens at OpenAI's own rates and appear in the model list. Where a request was overcharged, the audit log shows a price correction under it and the difference is credited to your spend and budgets.
- Changed
Fair pricing for models without a known price
A model whose price we cannot yet confirm is billed at the typical price of the same provider's comparable models, never at the most expensive model we know. We are alerted at once and add its real price. We now also check prices against two independent sources after every update.
- New
Videos in the docs
A new Videos page in the docs shows Sluis in real use: how it works, a day with Workspace, an agent going live on Gateway, and full tours of the console and of Workspace. Every film has a voice-over and captions in your language, plays in 1080p or 1440p, and can be downloaded.
- Changed
Privacy Policy updated: which Sluis apps you use
Our Privacy Policy now says that Sluis records which app you open (web Console, web Workspace, macOS, Windows, Linux, iOS or Android), on which day, and its version. 7Lab uses this to support customers and improve Sluis, keeps it for 12 months, and deletes it with your organisation. It is not shared with anyone else. You are asked to accept the new policy once.
- Improved
Workspace filter in the audit log
A new Workspace filter shows only the requests made from Workspace chats. The cost column is wider, so amounts below a cent are no longer cut off, and the page drops two banners that repeated information shown elsewhere.
- Fixed
Images follow your document inspection
Images sent in a request now get the same checks as documents. With document inspection on, PNG and JPEG images are scanned, and personal details in them are masked or replaced according to your settings. With the block setting, an image that cannot be checked is refused; otherwise it is sent on and marked as unscanned in the audit log and in the response.
- Fixed
Cached answers show your own details
With pseudonymisation and the answer cache on, an answer served from the cache could show a placeholder such as «EMAIL_1» instead of the detail you sent. Cached answers now show your own details, just like fresh ones.
- Fixed
Agency two-factor rules reach client organisations
When an agency requires two-factor sign-in, its owners and admins now need it in the client organisations they manage as well, not only in the agency itself.
- Changed
A clearer Data Processing Agreement
Version 8 of the Data Processing Agreement spells out who is responsible for account data, how long cached answers, Workspace chats and the names in them are kept, and that the mapping behind pseudonymized names is never stored outside Workspace. Owners and admins are asked to accept it once.
