Today we’re releasing a powerful update to Lumo that gives you a more capable privacy-first AI assistant offering faster, more thorough answers with improved awareness of recent events.
Guided by feedback from our community, we’ve been busy upgrading our models and adding GPUs, which we’ll continue to do thanks to the support of our Lumo Plus subscribers. Lumo 1.1 performs significantly better across the board than the first version of Lumo, so you can now use it more effectively for a variety of use cases:
Get help planning projects that require multiple steps — it will break down larger goals into smaller tasks
Ask complex questions and get more nuanced answers
Generate better code — Lumo is better at understanding your requests
Research current events or niche topics with better accuracy and fewer hallucinations thanks to improved web search
The latest upgrade brings more accurate responses with significantly less need for corrections or follow-up questions. Lumo now handles complex requests much more reliably and delivers the precise results you’re looking for.
In testing, Lumo’s performance has increased across several metrics:
Context: 170% improvement in context understanding so it can accurately answer questions based on your documents and data
Coding: 40% better ability to understand requests and generate correct code
Reasoning: Over 200% improvement in planning tasks, choosing the right tools such as web search, and working through complex multi-step problems
Most importantly, Lumo does all of this while respecting the confidentiality of your chats. Unlike every major AI platform, Lumo is open source and built to be private by design. It doesn’t keep any record of your chats, and your conversation history is secured with zero-access encryption so nobody else can see it and your data is never used to train the models. Lumo is the only AI where your conversations are actually private.
Unlike Big Tech AIs that spy on you, Lumo is an open source application that exclusively runs open source models. Open source is especially important in AI because it confirms that the applications and models are not being used nefariously to manipulate responses to fit a political narrative or secretly leak data. While the Lumo web client is already open source(new window), today we are also releasing the code for the mobile apps(new window). In line with Lumo being the most transparent and private AI, we have also published the Lumo security model so you can see how Lumo’s zero access encryption works and why nobody, not even Proton can access your conversation history.
