Ente, the privacy-focused cloud storage company, has released Ensu, an open-source application that runs large language models entirely on users’ devices without internet connectivity or data sharing. The app represents Ente’s first foray into local AI, building on the company’s previous work delivering on-device face recognition and image search in their photo storage service.
“LLMs are too important to be left to big tech,” the company stated in their announcement. Ente argues that while there remains a capability gap between frontier models and local alternatives, smaller models are improving rapidly and will soon cross a threshold where they become “good enough for most purposes” while providing “full privacy and control.”
Ensu functions as a ChatGPT-like interface but processes all queries locally using on-device models. The application supports image attachments and is available across iOS, Android, macOS, Linux, and Windows platforms, with an experimental web version. The core logic is written in Rust with native implementations for each platform. Users can engage in conversations about books, introspect on sensitive topics, and use the app offline during flights.
The release comes as an “Ente Labs project” focused on product iteration rather than immediate commercialization. The company acknowledges that Ensu “is currently not as powerful as ChatGPT or Claude” but describes the current version as “quite fun.” Future updates will include chat backup and synchronization across devices through Ente accounts with end-to-end encryption.
Ente plans to continue developing Ensu as local model capabilities improve, with the source code available for community contributions. The company positions this release as “just a checkpoint” rather than a final product, indicating ongoing development toward more capable local AI applications.
