Proton's Lumo 2.0 adds image generation, memory, thinking mode
TL;DR
- Proton released Lumo 2.0 on June 30, 2026, adding image generation and editing, a thinking mode, and user-controlled persistent memory within Projects.
- The company says Lumo 2.0 responds up to 76% faster than the prior version and preserves zero-access encryption with no server-side session logging.
- Proton claims Lumo 2.0 Lite scores 127% higher than Lumo 1.4 on the Artificial Analysis Intelligence Index, with Lumo 2.0 Max scoring 240% higher.
Privacy-first chat tools used to have an unglamorous problem: they were noticeably worse than the mainstream option, so most people compromised and used ChatGPT anyway. Proton is trying to change that equation with Lumo 2.0, and TechCrunch reports the update went live on June 30, 2026. It adds three things a mainstream user actually notices: image generation and editing, a thinking mode for harder questions, and user-controlled persistent memory scoped to Projects.
The headline performance claim is that Lumo 2.0 responds up to 76% faster than the previous version. Proton also points to the Artificial Analysis Intelligence Index, where it says Lumo 2.0 Lite scores 127% higher than Lumo 1.4 and Lumo 2.0 Max scores 240% higher. Those numbers are the company's, not an independent benchmark, so take them as marketing signal rather than settled fact. They still describe a serious version bump rather than a cosmetic one.
Founder and CEO Andy Yen framed the pitch bluntly, saying "Lumo 2.0 has been re-engineered from the ground up and the introduction of thinking mode gives it powerful new capabilities," and, in the same announcement, that users "no longer need to choose between powerful AI capabilities and meaningful privacy protections." The underlying story stays the same: zero-access encryption in transit and at rest, no server-side session logging, and a pledge never to use customer data for AI training or share it with third parties. Images and memory inherit those guarantees, which matters because both categories are exactly where mainstream chatbots typically retain the most.
The honest caveat is what the reporting does not give you. There is no independent benchmark of Lumo against ChatGPT, Claude or Gemini in this rollout, no detail on which underlying models actually run behind the Lite and Max tiers, and no operational specifics on how memory can be inspected, edited, or exported. Persistent memory, even user-controlled, is new attack surface: a Proton account compromise now leaks more than a stateless chat would.
Still, this matters because Proton is the closest thing the privacy camp has to a credible full-stack contender, and it now sells Lumo alongside a free tier through Plus and Professional plans. If Lumo 2.0 lands close to the mainstream in day-to-day quality, the no-training alternative becomes viable for people who had been quietly using ChatGPT with a thumb over the sensitive parts. That group, and the enterprises quietly worried about their data policy, is the one worth watching over the next quarter.
Originally reported by techcrunch.com
Read the original article →Original headline: Proton Launches Lumo 2.0 — Adds Image Generation, Thinking Mode, and Persistent Memory to Zero-Access Encrypted AI Chatbot, Claims 76% Faster Responses