v2.1.0

[github-actions[bot]]

📦 Open-WebUI-Functions v2.1.0

Warning

Parts of these release notes were generated with AI and may contain mistakes or inaccuracies. Please verify important details against the actual commits, pull requests, and source changes.

Note

Google Gemini 1.20.0 for Open WebUI 0.12:

• Tool Search;
• context compaction;
• chats shared with "Allow replies";
• thinking levels per Gemini 3 model.

No breaking changes, and no new valves. All other functions are unchanged since v2.0.0. Tested with Open WebUI 0.11.3, 0.11.4 and 0.12.0 (end-to-end suites against provider mocks).

This release updates the Google Gemini pipeline to 1.20.0. It now works with the Open WebUI 0.12 features that reach the model: Tool Search, context compaction and chats shared with "Allow replies". It also sends each Gemini 3 model the thinking levels it supports.

✨ New Features

• Chats shared with "Allow replies" (Open WebUI 0.12): participants who write in a shared chat now get the chat's earlier generated images and the owner's uploads for image edits. Generated images are linked to the chat, as with Open WebUI's own image generation, so participants no longer see them as broken (404). Access follows Open WebUI's own checks for continuing a chat and reading a file. On Open WebUI 0.11 the owner/admin rule is unchanged.
• Tool Search (Open WebUI 0.12): Gemini can call a tool that it found through search_tools. Before, the call was rejected because Open WebUI leaves deferred tools out of the request.

🐛 Fixes

• Context compaction: when Open WebUI's summary request to Gemini fails or returns nothing, the pipe returns an empty answer instead of its error text. Open WebUI then uses its own summary instead of storing an error message as the conversation summary (affected Open WebUI 0.11 too).

• Thinking levels per Gemini 3 model, following Google's docs:

  • gemini-3-flash, 3.5-flash-lite and 3.6-flash: minimal to high;
  • gemini-3.1-pro and 3.8-flash: low, medium and high;
  • gemini-3-pro: low and high.

  Before, every Gemini 3 text model was limited to low/high, so medium became high.

• Invalid thinking parameters: a non-string reasoning_effort or a non-numeric thinking_budget is ignored with a warning instead of dropping the whole thinking configuration. A number sent as a string ("1024") is accepted.

🔒 Security

• Files named by another participant's message in a shared chat: such a file is read only if that message's author may read it too. This prevents a participant from getting an image derived from the owner's private file.
• Smaller fixes from the security review, all low severity: the chat is checked again before an image is linked to it, the number of history images checked per request is capped, and the chat access check accepts only the chat itself.

⚠️ Known limits

• Images generated before Gemini 1.20.0 are not linked to their chat, so participants of a shared chat still cannot see them.
• Not verified against the real Gemini API: that it rejects undeclared tool calls, and that it accepts a turn whose earlier tool rounds were compacted. The end-to-end mocks model both.

🏗️ Infrastructure & CI/CD

• New Gemini end-to-end groups for shared chats and the Open WebUI 0.12 features. The default test image is now v0.12.0-slim: 580 checks pass on 0.12.0, and 565 each on 0.11.4 and 0.11.3.
• The static Open WebUI API check recognizes submodule imports. A missing method that a function looks up with getattr first is reported as a warning.

---

📋 Release Information

• Application: Open-WebUI-Functions
• Version: 2.1.0
• Repository: owndev/Open-WebUI-Functions
• Commit: 26399d912af9e89964c9faece17814e509414c04
• Release Type: ✅ Stable Release
• Environment: Production
• Created: 2026-10-11T22:45:18+02:00

📋 Version Information

• Full Version: 2.1.0
• Assembly Version: 2.1.0.60000
• Informational Version: 2.1.0+Branch.main.Sha.26399d912af9e89964c9faece17814e509414c04
• Pre-release Tag: None
• Pre-release Number: 0
• Commits Since Version: 0
• Weighted Pre-release Number: 60000

🛠️ Build Information

• Workflow: 🏷️ GitHub Release
• Run ID: 38173912295
• Actor: @owndev

🐛 Support & Feedback

📋 Need Help?	🔗 Resource
🐛 Report Issues	owndev/Open-WebUI-Functions/issues
💬 Discussions	owndev/Open-WebUI-Functions/discussions
📚 Documentation	owndev/Open-WebUI-Functions/wiki

What's Changed

• Gemini 1.20.0: Open WebUI 0.12 features, shared chats, thinking levels per model by @owndev in Gemini 1.20.0: Open WebUI 0.12 features, shared chats, thinking levels per model #201
• E2E: default image Open WebUI v0.12.0-slim by @owndev in E2E: default image Open WebUI v0.12.0-slim #202

Full Changelog: v2.0.0...v2.1.0

---

This discussion was created from the release v2.1.0.