If you haven’t checked out the Open WebUI Github recently, now is the perfect time! The development team has been on a roll, introducing groundbreaking features that are transforming how users interact with language models. From video chatting with vision-capable models to a sophisticated tool library, there’s a lot to explore.
One of the most exciting new features is the LLM video chatting capability. You can now open your camera and ask your model questions like, “how many fingers am I holding up?” and get a response! This feature leverages vision-capable models such as Moondream and LLava, and impressively, all TTS and STT processes are handled locally. Just make sure you’re serving your Open WebUI over SSL for optimal performance.
Another notable addition is the tool library and function calling feature. While the documentation might still be catching up, this feature offers functionality similar to what Autogen and Crew AI provide. It can be found under Workspace > Tools, making it easier to enhance your models with additional tools.
The update also includes per model knowledge libraries, allowing users to upload PDFs and enrich the model’s knowledge base on specific topics. This feature, accessible at the bottom of each model’s config page, enables more domain-specific customization, akin to GPT4ALL’s content libraries.
Lastly, the ability to run generated Python code directly in chat is a game-changer, though it comes with security risks. This feature is perfect for quickly testing functions before moving them to more robust environments like VS Code. Simply click the “run code” link in the top-right corner when your model generates Python code.
These updates are just the tip of the iceberg. The Open WebUI team’s relentless innovation is not just about adding new features but about significantly enhancing user experience. Dive into their Github and explore these incredible advancements today!
