Proton's Lumo 1.1 Proves Powerful AI Needs No Data Sacrifice

In a significant move to challenge the dominance of data-hungry artificial intelligence, Swiss privacy-focused company Proton has rolled out a major upgrade to its AI assistant, Lumo. The new version, Lumo 1.1, delivers substantial enhancements in speed, intelligence, and accuracy, aiming to prove that a powerful AI tool does not require the sacrifice of personal privacy.[1][2] This update sharpens Lumo's capabilities across a wide range of tasks, positioning it as an increasingly viable alternative to mainstream assistants from tech giants, while holding firm to its foundational commitment to user confidentiality. The core of the upgrade lies in a dramatic performance boost, making the assistant more capable and reliable for complex user requests.[1]

The enhancements in Lumo 1.1 are not minor tweaks but a considerable leap forward in functionality. Proton claims the updated assistant demonstrates a 170 percent improvement in understanding the context of a user's query, leading to more accurate and relevant answers.[1][3] For developers and programmers, the assistant now boasts a 40 percent improvement in generating and assisting with software code.[1][2] The most significant reported gain is an over 200 percent improvement in planning, reasoning through complex multi-step problems, and choosing the right tools, like its web search function, to tackle a prompt.[1][3] This means the AI is now better equipped to handle intricate tasks, from planning projects to researching current events, an area where the initial version was perceived to lag behind competitors.[1][3] These improvements are the result of upgrading the underlying models and adding more processing power through GPUs, an expansion directly supported by subscribers to the premium Lumo Plus service.[1]

Despite these substantial performance upgrades, the centerpiece of the Lumo offering remains its steadfast, privacy-first architecture, a stark contrast to the standard operating procedures of many large-scale AI providers.[4][5] Proton has built Lumo on a "zero-logs" policy, meaning that user conversations are not stored on its servers by default.[4][6] For users who wish to save their chat history, the data is protected by zero-access encryption, a security measure ensuring that only the user, and not even Proton, can decrypt and read the conversations.[6][5][7] Crucially, the company guarantees that user inputs are never used to train its AI models, a common practice among other AI services that has raised significant privacy concerns.[4][6] This prevents sensitive or personal information shared in a chat from appearing in another user's query results. To bolster these claims with verifiable proof, Proton is also making the source code for Lumo's mobile applications open-source, allowing the security community to audit its privacy and security credentials.[3][2]

Underpinning the new Lumo 1.1 is a strategic reliance on powerful open-source large language models (LLMs). Proton has disclosed that its assistant is now powered by a selection of models including Mistral Small 3, Nemo, OLMO 2 32B, and OpenHands 32B.[1][6] These models are hosted exclusively on Proton's own servers within its European data centers, a strategy the company refers to as its "Eurostack."[1][5] This approach is designed to provide greater transparency and protect users from the broad surveillance reach of governments in other parts of the world.[1][5] By avoiding partnerships with major American or Chinese AI firms and running its own infrastructure, Proton maintains full control over the data pipeline, ensuring that user queries are never sent to or processed by any third parties.[6][5] This self-reliant, open-source strategy is a fundamental departure from models like Apple's "Private Cloud Compute," which involves partnerships with external AI providers.[5]

In conclusion, the launch of Lumo 1.1 marks a critical juncture for both Proton and the broader AI industry. By significantly closing the performance gap with its competitors while reinforcing its uncompromising privacy guarantees, Proton is issuing a direct challenge to the notion that AI advancement must be fueled by user data. The upgrade moves Lumo from being simply a private alternative to a private and powerful one. The success of this model, which is funded by direct user subscriptions rather than by monetizing data, could signal a growing market appetite for technology that respects user autonomy.[2][8] As users become increasingly aware of how their digital footprints are being used, Lumo's dual emphasis on capability and confidentiality may set a new standard, pressuring the industry's giants to reconsider the ethical and privacy implications of their own AI development.[5][2]