The documentation has two settings fields reversed when compared to Open WebUI actual settings screen. This causes some people (like me) to input the model in the voice box and the other way around, causing strange errors. By swapping the two fields in the docs, we follow natural order of information (top to bottom, left to right) and make the documentation match the actual settings screen.
Replaces the existing `models.md` file with a comprehensive new version.
This updated documentation details the Open WebUI Preset System, explaining how to wrap base models to create custom agents with specific system prompts, knowledge collections, tools, and dynamic variables. It covers creation, management, and advanced parameters.
Introduces a new, comprehensive documentation page for the "Notes" feature.
This page explains the core functionality of the Notes editor, its AI integrations (Enhance, Chat sidebar), and how notes can be used to provide deterministic context in standard chats. It also details use cases and limitations to provide a complete understanding of the feature.
Introduces a new documentation page for the "Channels" feature.
This page explains what Channels are, how to enable and use them, and provides several use cases for this new collaborative tool. It details the unique `@mention` system for interacting with different AI models within a shared timeline.
Adds several key clarifications to the Gemini image generation page:
- A warning that a payment method is required for paid API keys.
- A tip that a Google Cloud API key from Vertex AI can be used as an alternative to a service account.
- A minimal working setup example to guide users in their configuration.
Adds a prominent warning to the tool development documentation regarding the potential for package version conflicts when multiple tools are installed with conflicting dependencies.
The warning recommends using OpenAPI tool servers as the robust solution to this problem, as it isolates tool dependencies.
