AnythingLLM and You.com bring real-time web search to everyone
You.com's accurate, fresh, and fast web search is now the default in AnythingLLM. Web search is on from your first chat, with no API key and nothing to setup.
September 2026
The web is the largest store of human knowledge ever built. Every fact, every release note, every forum thread where someone already solved your problem is out there. A model that cannot reach it is working from memory alone, and memory has a cutoff.
The moment a model needs something it was not trained on, whether that is help, context, or just what happened this morning, giving it real access to the web is paramount. The model can only be as good as what it can find, and the quality of the search behind it decides the quality of the answer.
Today we are excited to announce a collaboration with You.com, the leading web search API for AI. Starting now, You.com's accurate, fresh, and fast web search is the default search provider in AnythingLLM Desktop, Mobile, and on the self-hosted server. Millions of AnythingLLM users get real-time web search out of the box, with no account, no API key, and nothing to configure.
Search is everything
Model's knowledge of the world events or information only extend to the day training stopped. Ask it about something that happened last week and it will either tell you it does not know, or worse, it will guess.
This is doubly true for local models. The models most AnythingLLM users run on their own hardware have fixed weights. They are frozen at a point in time, and the world keeps moving. A local model with no connection to the web is a very smart person who has been in a room with no windows for a year.
The fix is giving the model you already have a live, reliable way to look things up. Real-time web search, retrieval, and deep research all rest on the same foundation, a strong search provider that returns results that are accurate, current, and rich enough for a model to actually use. Weak search means weak answers or outdated ones. It really is that simple.
We have supported web search in AnythingLLM agents for a long time, across a dozen providers with DuckDuckGo being the default. Web search was also opt-in and not a default tool so some people never even thought to turn it on.
Why You.com
You.com builds the real-time web data layer for AI. Their search APIs are designed for agents and LLMs from the ground up, with results that are structured, clean, and ready to drop into a prompt, and they back it up with the top score on the SimpleQA benchmark and peer-reviewed evaluation research. They are focused on exactly the three things that matter when a model is the one doing the reading: accuracy, freshness, and speed.
Just as important, You.com shares our stance on privacy. The endpoint AnythingLLM uses is zero data retention. Your searches are not stored, not logged against you, and not used to train anything. That is a perfect match for what we want AnythingLLM to be. We build the experience. You.com delivers the web on request.
What this means for you
If you are new to AnythingLLM, web search is on from your very first chat. Install, ask a question about something that happened this morning, and your agent goes and finds out. You.com is the default provider and there is nothing to set up.
If you already use AnythingLLM, web search is now enabled by default the next time you update, and You.com is the provider unless you have already configured one. If you set up your own search provider before, nothing changes for you. If you never got around to it, you are about to get a lot more out of your assistant.
Richer results. We now pass every snippet You.com returns, along with published dates, into your agent's context. Your model sees more of the page and knows how old it is, which means better answers and fewer confident mistakes.
It always works. If a You.com request fails or comes back empty for any reason, AnythingLLM falls back to DuckDuckGo automatically. Your search never silently breaks.
Desktop, Mobile, and server, both. This is the same experience whether you run AnythingLLM Desktop on your laptop, your Android device, or the self-hosted server for your whole team.
Zero data retention. AnythingLLM has always been built on the promise that your data is yours. Web search is no exception. The You.com endpoint we use retains nothing, so your queries are not stored or used to train anything. The searches your assistant runs stay as private as the rest of your workspace.
Admins stay in control. If you have turned web search off for your instance, it will stay disabled.
Try it
- Download AnythingLLM Desktop or update to the latest version. Server users can pull the latest Docker image.
- Start a chat and ask about anything happening right now.
- That is it. Your model just got a window to the outside world complete with citations.
Thank you, You.com
A huge thank you to the team at You.com for partnering with us on this. Our mission is to make powerful, private AI something anyone can use, and search that is accurate and current is a huge part of that. Bringing You.com to every AnythingLLM user is a big step, and we are just getting started.
The default You.com does apply reasonable rate-limiting per day per network. If you need more, you can add your own You.com API key in the web search settings and keep going or rely on our fallback search results. Questions or feedback? Find us on GitHub.
Learn more