Ente Releases Ensu: A Privacy-First Local LLM App That Runs Entirely On-Device

Ensu: Local LLMs with Full Privacy, Zero Cost

Ente, the company known for its encrypted photo storage app, has released Ensu - a ChatGPT-like app that runs entirely on-device with full privacy and zero recurring cost.

Why Local LLMs Matter

Ente's founders argue that "LLMs are too important to be left to big tech." Their concerns include:

• Privacy - Conversations with centralized LLMs are not truly private
• Dependency - Arbitrary bans, content shaping, and non-portable memory
• Manipulation risk - LLMs can be used to manipulate people en masse
• Control - Users should own their AI, not rent it

What Ensu Does

• Runs completely offline on your device - no internet required
• Cross-platform - iOS, Android, macOS, Linux, Windows, and experimental web
• Open source - Built on the Ente open source framework (MIT/AGPL)
• Core logic in Rust - Native mobile apps and Tauri desktop apps
• Image attachments supported - Multimodal capabilities
• End-to-end encrypted sync coming - Chat backup across devices via Ente account or self-hosting

The Capability Gap

Ente acknowledges that local models aren't yet as powerful as GPT-4 or Claude. But they draw a parallel to their earlier work on local photo ML features (face recognition, person clustering, NL image search) which were once thought impossible on-device but are now standard.

Use Cases

• Private introspection about thoughts you wouldn't risk in non-private LLMs
• Reading and discussing books (works surprisingly well with classics)
• Using on flights or places without internet
• Any sensitive professional or personal conversations

The app is available for download now at ente.com/ensu and is an Ente Labs project - still experimental but already quite usable.