When a group change affects its members (new permissions, parent group or default models, or deleting the group), the server makes every member's browser reconnect so it picks up the new permissions without a page reload. To find each member's connections it read through the whole list of connected users again for every single member, so saving a 2,000-member group with 3,000 people online spent almost 9 seconds on this with Redis as the websocket manager, and about 1 second without. It now finds all members' connections in one pass.
Changing one permission on a 2,000-member group, time the save spends making members reconnect:
| Online users | Redis, before | Redis, after | No Redis, before | No Redis, after |
| --- | --- | --- | --- | --- |
| 1,000 | 2.8 s | 0.15 s | 0.09 s | <0.01 s |
| 3,000 | 8.7 s | 0.28 s | 1.2 s | <0.01 s |
In the PowerPoint preview, a title or subtitle that the file sets to left alignment was drawn centred, the same as one with no alignment at all. Decks made by python-pptx and most deck generators set the alignment on every paragraph, so their left-aligned title slides all came out centred. The preview now keeps a title the file aligns left on the left, and only centres titles and subtitles that leave the alignment open.
Fixes#32873
The web UI still shipped version 3.4.11 of DOMPurify, the library that strips unsafe HTML before content is shown, six releases behind DOMPurify's own latest. This moves it to 3.4.17, and new installs now require at least that version.
Co-authored-by: wesyoung2022 <115102004+wesyoung2022@users.noreply.github.com>
Importing a model JSON on the Workspace Models page as a regular user skipped any model whose ID already belonged to another user's model they cannot edit, or to a model from one of the instance's connections such as OpenAI or Ollama. There was no message, so the model just never showed up. The import now stops at that model with the error "Uh-oh! This model id is already registered. Please choose another model id string."; the models before it in the file are imported and the ones after it are not. Admins, and users who can edit the existing model, are unaffected and still overwrite it.
The Docker image now runs Python 3.12 on Debian 13 (trixie) instead of Python 3.11 on Debian 12, and the pip install docs recommend 3.12. Image size stays about the same: the slim image is 1 MB smaller and the standard image is about 50 MB larger, mostly from the newer ffmpeg and pandoc that Debian 13 ships. Admins should know that custom functions, tools and anything installed with pip inside the container now run on Python 3.12, so a package that only has builds for 3.11 needs a newer version. Installing with pip still works on both Python 3.11 and 3.12.
Picking a model control option, such as a Thinking level, needed both the Allow Chat Controls and Allow Chat Params permissions, so admins had to hand out the Advanced Parameters panel just to let people pick a preset. Model control picks now work for anyone with access to the model and are saved per model to their account. Allow Chat Controls still hides the Chat Controls panel and Allow Chat Params still hides the Advanced Parameters panel, in the chat and in personal Settings. Parameters a person saves to their account are no longer filtered by these two permissions, since any chat message could already set any parameter without them.
Fixes#32626
A self-hosted MinerU 4.x server started with an API key rejected every document with a 401, even with the key filled in under Documents settings, because local mode never sent it. The key is now sent with the document requests to a 4.x server whenever one is set. Servers without a key are unaffected.
Fixes#32836
A model connection with its auth type set to OAuth only forwarded the user's sign-in token for requests coming from the browser. The same user calling the API with their bearer token or an API key sent nothing upstream, so an SSO-protected model server rejected every API call and its models could not be listed or used. These connections now fall back to the user's latest sign-in session the same way tool servers already do.
Fixes#32820
Sending the first message while a freshly loaded page was still setting up the new chat wiped that message from the screen once the setup finished. The reply then stayed empty with the Stop button on, or the page fell back to an empty new chat while the finished chat only showed up in the sidebar. The setup now leaves the chat alone when a message was sent in the meantime.
Fixes#32819
With Calendar switched off by an admin, or the calendar permission taken away, the sidebar entry was hidden but opening /calendar directly from a bookmark or the address bar still showed an empty month grid with a "401 Unauthorized" error. The page now sends these users to the home page, the same as Notes and Automations do.
Fixes#32834
A temporary chat offered Fork chat under every reply and Fork in the / menu, but a temporary chat is never stored, so there was nothing to copy and every attempt ended in "We could not find what you're looking for :/". Both are now left out of temporary chats, the same way rating already is, and typing /fork there answers with "No chat to fork". Saved chats fork as before.
Fixes#32833
With web search off, clicking Save on the Web Search settings page jumped to an empty search engine field and saved nothing. That happened with no engine picked, which is how a fresh install starts, and also when an engine picked earlier was missing its own details, such as the SearXNG query URL. Admins could not change the Web Loader settings, which apply whenever a user attaches a web page, without filling in a search engine first. The engine and its details are now only required while web search is turned on.
Fixes#32831
With Allow Chat Edit switched off, users lost the Edit button on replies but kept it on their own messages, and both Save and Send still worked there. The Edit button on a user's own messages now follows the same permission, as it already did in chats shared with replies allowed. Admins keep it as before.
Fixes#32832
With OAuth 2.1 (Static), when the authorization server did not accept the client secret as a form field, Open WebUI used whichever auth method the server's metadata listed first. X Ads lists "none" (public client, no secret) first, so the token exchange went out without the configured client secret and X rejected it with "Missing valid authorization header". When the server supports HTTP Basic auth, the secret is now sent that way, no matter which method it lists first.
Fixes#32813
Holding Shift while deleting a file in Manage Files used to delete it at once, but since the redesign every delete asks "Delete 1 file?" first. Shift+click or Shift+Enter on a file's Delete option now deletes it straight away again, and the option turns red while Shift is held, as before. Delete without Shift still asks.
Fixes#32814
The number next to the Files title in Manage Files stayed at the account's total while a search narrowed the list. While searching it now shows how many files are listed for the search, as before the redesign, and the total comes back once the search is cleared.
Fixes#32815
Create and publish Docker images with specific build args / build (map[arch:linux/amd64 runner:ubuntu-latest], map[build_args:USE_CUDA=true
USE_CUDA_VER=cu126
free_disk:true name:cuda126 suffix:-cuda126]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/arm64 runner:ubuntu-24.04-arm], map[build_args:USE_CUDA=true
USE_CUDA_VER=cu126
free_disk:true name:cuda126 suffix:-cuda126]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/arm64 runner:ubuntu-24.04-arm], map[build_args:USE_CUDA=true free_disk:true name:cuda suffix:-cuda]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/arm64 runner:ubuntu-24.04-arm], map[build_args:USE_OLLAMA=true free_disk:false name:ollama suffix:-ollama]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/arm64 runner:ubuntu-24.04-arm], map[build_args:USE_SLIM=true free_disk:false name:slim suffix:-slim]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/amd64 runner:ubuntu-latest], map[build_args: free_disk:false name:main suffix:]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/amd64 runner:ubuntu-latest], map[build_args:USE_CUDA=true free_disk:true name:cuda suffix:-cuda]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/amd64 runner:ubuntu-latest], map[build_args:USE_OLLAMA=true free_disk:false name:ollama suffix:-ollama]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/amd64 runner:ubuntu-latest], map[build_args:USE_SLIM=true free_disk:false name:slim suffix:-slim]) (push) Waiting to run
Create and publish Docker images with specific build args / build (map[arch:linux/arm64 runner:ubuntu-24.04-arm], map[build_args: free_disk:false name:main suffix:]) (push) Waiting to run
After reloading a chat while a background sub-agent was still working, its report came in as a plain user bubble with the whole raw report text instead of the "Background sub-agent finished" row, until the next reload. When the report arrives the page reloads the chat and receives the new turn live at the same moment, and the live copy, which lacks the message details, replaced the complete one from the server. The live copy is now laid over the server copy, so live changes still win and the report keeps what marks it as a sub-agent report.
Fixes#32801
