Lumo 2.0's benchmark jump, Proton says its Lite tier scores 127% higher and Max scores 240% higher than Lumo 1.4 on the independent Artificial Analysis Intelligence Index, is the more consequential detail for practitioners than the image-generation headline, because it suggests privacy-first assistants are starting to close a real capability gap with mainstream models rather than just adding surface features.
What happened
Proton released Lumo 2.0 on June 30, adding image generation and editing, a "thinking" reasoning mode alongside a faster default mode, persistent memory, and private web search, according to Proton's own announcement and TechCrunch. Eamonn Maguire, Proton's Director of Engineering for ML and AI, described it as the product's largest capability leap yet. All image processing, like Lumo's existing chat data, is protected by zero-access encryption, which Proton says means it cannot access user content itself. Lifehacker published a hands-on review comparing the updated Lumo to ChatGPT; Proton first launched Lumo in August 2025 and continues to market it as keeping no logs and not training on user conversations.
Technical context
Proton says Lumo 2.0 Lite scores 127% higher and Lumo 2.0 Max scores 240% higher than Lumo 1.4 on the Artificial Analysis Intelligence Index, an independent benchmark aggregating tests including GDPval-AA v2, Terminal-Bench v2.1, Humanity's Last Exam, and GPQA Diamond; Proton also says everyday queries now respond up to 76% faster. These are vendor-reported benchmark results rather than figures independently reproduced by Artificial Analysis or a third party in this reporting, so they should be read as Proton's characterization pending outside verification.
For practitioners
Privacy-first assistants change deployment trade-offs for consumer-facing or regulated applications by altering data-handling constraints without requiring self-hosted models. Lumo 2.0's move toward feature parity, including image generation, reasoning modes, and memory, matters for feature decisions like multimodal input support and context handling, and for compliance workflows where zero-access encryption and no-training guarantees can lower regulatory risk.
What to watch
Watch for independent, third-party validation of the Artificial Analysis Intelligence Index gains, documented APIs or SDKs enabling integration beyond Proton's own apps, and how Lumo Plus pricing ($10/month for unlimited chats and advanced image generation) compares to mainstream assistant subscriptions as adoption data emerges.
Key Points
- 1Proton released Lumo 2.0 on June 30, 2026, adding image generation, a reasoning mode, memory, and private web search to its encrypted privacy chatbot.
- 2Proton claims Lumo 2.0 scores 127-240% higher than the prior version on an independent AI benchmark, though results aren't yet third-party verified.
- 3Privacy-preserving assistants gaining mainstream features shift practitioner trade-offs around compliance, data handling, and integration for regulated AI deployments.
Scoring Rationale
A well-covered privacy-focused chatbot update with real capability additions (image generation, reasoning mode, memory) and a notable vendor-reported benchmark jump versus the prior version, corroborated across multiple independent outlets. Benchmark claims remain vendor-sourced pending third-party reproduction, and this is a product update rather than a new foundation model, keeping it in the 'solid' tier.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
