product update

Proton launches Lumo 2.0 with multimodal capabilities, scores 240% higher on AI benchmarks

TL;DR

Proton has released Lumo 2.0, adding image recognition and generation, encrypted memory features, and enhanced web search to its privacy-focused AI assistant. The company claims Lumo 2.0 Max scored 240% higher than version 1.4 on the Artificial Analysis Intelligence Index, while maintaining zero-access encryption and no conversation logging.

2 min read
0

Proton launches Lumo 2.0 with multimodal capabilities, scores 240% higher on AI benchmarks

Proton has released Lumo 2.0, adding image recognition and generation, encrypted memory features, and enhanced web search to its privacy-focused AI assistant. The company claims Lumo 2.0 Max scored 240% higher than version 1.4 on the Artificial Analysis Intelligence Index.

Performance gains and new features

Lumo 2.0 Max scored 240% higher than Lumo 1.4 on the Artificial Analysis Intelligence Index, according to Proton. The index combines results from nine benchmarks including GDPval-AA v2, Terminal-Bench v2.1, Humanity's Last Exam, and GPQA Diamond. Lumo 2.0 Lite scored 127% higher than version 1.4 on the same index.

The update introduces three major capabilities:

  • Image recognition and generation: Multimodal features allowing users to analyze, edit, and generate images within conversations, protected by zero-access encryption
  • Memory and context: User-controlled memory, encrypted Projects feature, and Custom Lumos for personalized workflows
  • Web search: Enhanced search with live results and source citations

Proton maintains its core privacy principles: no logs, zero-access encryption, no data sharing with third parties, no use of conversations for AI training, and reliance on open-source language models.

Business offering and adoption

Proton claims more than 10 million people have adopted Lumo as a privacy-focused alternative to mainstream AI platforms. The company is targeting businesses with Lumo for Business, emphasizing that conversations are zero-access encrypted, never logged, and never used for model training. The service runs on independent European infrastructure, which Proton says protects it from US Executive Orders and American data collection requests.

Pricing and availability

Lumo 2.0 is available now with three tiers:

  • Free tier: Core AI features for everyday private use
  • Lumo Plus: Unlimited chats, Projects, advanced image generation, and access to Proton's most capable models (pricing not disclosed)
  • Lumo Professional: Advanced capabilities for teams requiring secure AI collaboration (pricing not disclosed)

What this means

Proton is positioning Lumo 2.0 as direct competition to ChatGPT and Claude, claiming user testing shows users can no longer perceive qualitative differences for many use cases. The 240% benchmark improvement is significant, though the specific scores and how they compare to OpenAI and Anthropic's latest models remain unclear. For organizations handling sensitive data, the combination of European infrastructure and zero-access encryption may offer a compelling alternative to US-based AI services, particularly given recent executive actions affecting AI data governance.

Related Articles

product update

Google brings personalized image generation to all US Gemini users, expanding from paid-only feature

Google is expanding personalized image generation in the Gemini app to all eligible US users, removing the previous restriction to AI Pro and Ultra subscribers. The feature allows Gemini to access user data across Google services like Gmail and Photos when generating images.

product update

Google makes Gemini's personalized image generation free for all U.S. users

Google removed the paywall for Gemini's personalized image generation feature, making it free for all eligible U.S. users starting today. The Nano Banana-powered feature was previously limited to Plus, Pro, and Ultra subscribers.

product update

AWS demonstrates two-model pipeline using Nova 2 Lite and Claude Sonnet 4.6 that cuts document processing costs by 67%

AWS published a technical demonstration showing that pairing Amazon Nova 2 Lite with Anthropic's Claude Sonnet 4.6 reduces document processing costs by approximately two-thirds compared to single-model approaches. The two-stage pipeline processed 336 scanned yearbook pages at $0.0027 per page, producing 3,122 name-to-face associations with 93% scoring at or above 0.95 confidence.

product update

Base44 launches Base1 LLM trained on tens of millions of user interactions

Base44, the vibe coding platform acquired by Wix for $80 million in 2025, has released Base1, a custom LLM trained on tens of millions of user interactions. The company claims the model will eventually outperform frontier models through optimization for latency, cost, and efficiency specific to app creation workflows.

Comments

Loading...