Proton Lumo Review

Lumo is a chat assistant from the team behind Proton Mail and Proton Drive. It offers image generation with refinement, web search, document analysis, projects, and code review, with the option to link Proton Drive files for context. Proton says the models, which include Qwen, GLM, and Apertus, are open source and run only on servers it controls. Stored chats use zero-access encryption so only the account holder can read them, and Proton says queries and responses are erased after processing and that Lumo keeps no metadata such as timestamps or IP addresses. Guests and Tor users can try it without an account.

There is nothing to install: Lumo runs in the browser and in Proton's apps, and you sign up for a Proton account, or use it as a guest. The open-source part is the client code, which sits in Proton's WebClients monorepo under GPL-3.0 (the star count here is for the whole monorepo, not only Lumo). The models run on Proton's servers, so you cannot self-host the service.

It follows the familiar chat layout, with projects to group related work, and Proton emphasizes that your history is saved only on your side. A free tier offers limited messages and image generations and one project, and Lumo Plus raises those limits and adds custom assistants. We did not test the service ourselves; this section reflects Proton's privacy documentation and OpenAlternative.

• Zero-access encryption for saved chat history
• No-logs policy and no training on conversations, according to Proton
• Image generation, web search, and document analysis
• Projects and optional Proton Drive file context
• Anonymous access as a guest, including over Tor

• Strong, clearly documented privacy model from a company known for it
• Swiss jurisdiction, GDPR compliance, and open-source models and clients
• Part of the Proton ecosystem, so Drive and account integration is simple

• Hosted only, so you cannot run your own instance
• Open models are generally behind the best proprietary ones for hard reasoning and coding tasks
• Free tier is limited, and some features, such as custom assistants, require the paid plan
• Privacy claims are Proton's own and rely on trust in its servers, even though the clients are open source

Lumo has Guest, Free, and Plus access. The free tier has limited messages, image generations, and one project, and Lumo Plus adds unlimited chats, higher model usage, more image generations, and unlimited projects. We could not confirm the current Plus price from the pages we reviewed, so check proton.me.

ChatGPT, Claude, and Gemini are the capability leaders, with stronger models, larger tool ecosystems, and more integrations. Lumo trades some of that capability for privacy: no logs, no training on your chats, and open models on Proton's own servers. If your work involves sensitive material and the tasks are ordinary writing, summarizing, and Q&A, Lumo is a reasonable choice, but for the hardest tasks the frontier models are better.

• Privacy-focused users and Proton customers who want an AI assistant without data harvesting
• People who handle sensitive documents and want encrypted chat history
• Users who want an open-model assistant without running one themselves

• Teams that need the best possible reasoning or coding performance
• Anyone who wants to self-host or choose their own model
• Users who need a large plugin and integration ecosystem