Proton Rolls Out Lumo 2.0: Privacy-First AI Now Adds Multimodal Capabilities and ‘Thinking Mode’
Table of Contents
The Privacy Pivot in the LLM Race
Proton, the Swiss-based company known for its uncompromising stance on encryption and digital sovereignty, is attempting to bridge the gap between high-utility artificial intelligence and strict user privacy. On Tuesday, the company unveiled Lumo 2.0, a significant overhaul of its public AI chatbot designed to challenge the dominance of data-hungry giants like OpenAI and Google.
While the first iteration of Lumo served as a proof-of-concept for private interaction, version 2.0 transforms the tool into a multimodal assistant. The most immediate shift is the introduction of image recognition and generation. Users can now upload visual data for Lumo to analyze or request original imagery via text-to-image prompts—functionality that brings Lumo closer to the feature sets of ChatGPT-4o and Gemini Pro.
Beyond Basic Chat: Persistent Memory and ‘Thinking Mode’
The upgrade extends deep into Proton’s ecosystem, specifically within ‘Projects.’ This widget integrates Lumo with Proton’s suite of productivity tools, including Proton Mail and Proton Drive. The standout addition here is user-controlled persistent memory. Unlike standard LLM sessions that reset or rely on opaque system-wide profiles, Lumo 2.0 allows users to define preferences that the AI recalls across different conversations, granting a personalized experience without sacrificing the user’s control over their data footprint.
Proton has also introduced a ‘thinking mode,’ a dedicated processing state for complex reasoning and multi-step problem solving. This suggests a shift toward a more deliberate inference process, likely aimed at reducing hallucinations in technical or analytical queries. To complement this, the company claims a massive performance boost, asserting that Lumo 2.0 responds to the majority of queries up to 76 percent faster than its predecessor.
“Lumo 2.0 has been re-engineered from the ground up and the introduction of thinking mode gives it powerful new capabilities,” said Andy Yen, Founder and CEO at Proton. “Lumo 2.0 demonstrates that users no longer need to choose between powerful AI capabilities and meaningful privacy protections.”
The Encryption Architecture: How it Differs
The central value proposition of Lumo isn’t just what it can do, but what Proton claims it cannot do. In a landscape where most AI models are trained on user prompts to improve future iterations, Proton employs a zero-access encryption architecture. This framework ensures that data is encrypted both in transit and at rest, with keys held exclusively by the user.
Crucially, Proton maintains that it performs no server-side logging of sessions. This means that even employees at Proton cannot view the contents of a user’s conversation. By explicitly promising that customer data will never be used for AI training or shared with third parties, Proton is positioning Lumo as the sanctuary for professionals and privacy advocates who find the terms of service for mainstream AI tools too invasive.
In practical testing, Lumo 2.0’s output remains competitive, mirroring the detail and contextual nuance of top-tier LLMs. While the free version provides a gateway to these features, Proton is leveraging its subscription model by offering ‘Plus’ and ‘Professional’ tiers, which provide increased resource quotas and expanded access for power users.
