Lumo 2.0 Brings Image Recognition and Persistent Memory to Proton's Privacy-First Chatbot

Proton, the privacy-focused productivity suite provider, has upgraded Lumo, its end-to-end encrypted AI chatbot, to version 2.0 with multimodal capabilities and performance improvements. According to TechCrunch, the release includes image recognition, image generation, persistent conversational memory, and a reasoning mode designed for complex queries—features that bring Lumo’s surface-level functionality closer to mainstream competitors while preserving its encryption-first architecture.

Multimodal Capabilities and Persistent Memory

Lumo 2.0 now supports image uploads for analysis and editing, matching capabilities found in ChatGPT and Gemini. The chatbot can also generate images from text prompts. TechCrunch reports that the update expands Proton’s Projects feature—a workspace widget for uploading documents and integrating with Proton’s email and cloud-storage products—by adding user-controlled persistent memory. This function allows Lumo to recall user preferences across separate conversational sessions, reducing friction for workflows that span multiple interactions.

Performance and Reasoning Capabilities

According to TechCrunch, Lumo 2.0 processes most queries 76% faster than the previous iteration. The release introduces a “thinking mode” for handling multi-step reasoning problems, though Proton has not detailed whether this mirrors OpenAI’s o1-style extended reasoning or adopts a different approach.

Proton CEO Andy Yen stated that the redesign was undertaken from the ground up and that Lumo 2.0 aims to demonstrate that users need not sacrifice privacy for capability. TechCrunch notes that in functional comparison—format, detail level, contextual awareness—Lumo performs roughly equivalent to Gemini and ChatGPT, signaling that privacy protections no longer force users into a capability trade-off.

Privacy Architecture Unchanged

Lumo’s differentiation hinges on its zero-access encryption model. TechCrunch reports that Proton encrypts all user data in transit and at rest, permits only the user to decrypt it, and retains no server-side session logs. The company commits to never training on customer conversations or selling data to third parties. These promises remain in effect in version 2.0, and represent the core claim Proton makes against mainstream chatbot providers, which typically reserve rights to use conversations for model improvement and internal analytics.

Pricing and Availability

Lumo 2.0 is available immediately in a free public tier, with paid Plus and Professional subscription tiers offering expanded resource allocation, according to TechCrunch.

Why This Matters

The Lumo 2.0 release tests whether privacy-first design can compete on feature parity rather than positioning as a privacy-only alternative. For users whose threat model includes distrust of data handling at OpenAI, Google, or Anthropic, Lumo’s multimodal support and reasoning mode may now eliminate the need to choose between encrypted conversations and advanced capabilities. However, Lumo’s adoption remains marginal compared to ChatGPT or Gemini—the real test is whether persistent memory and image features catalyze migration from privacy-unconcerned users, or simply retain the small existing cohort of privacy-conscious adopters. If Proton can demonstrate that Lumo 2.0 meets 90%+ of mainstream use cases without exposing metadata or training data, it positions encryption-by-default as competitive rather than niche.