LM Pocket
Run local AI models on-device
About
Run open-source AI models directly on your device without cloud reliance. Discover, download, and chat with thousands of GGUF models privately. Offers a local API server and advanced management tools for power users.
What's New in LM Pocket
1.0.6
August 31, 2026
- Exact device-based model compatibility checks - Enhanced reliability for local API and streaming - Improved model loading and server controls - Persistent request logs and stronger validation - Per-model system prompts - Streamlined settings and model-first server experience
MoreFAQ
Does LM Pocket have ads?
No, LM Pocket is 100% free and does not contain any advertisements.
What AI models can I run with LM Pocket?
LM Pocket supports any GGUF-quantized model, including popular ones like Llama, Mistral, Phi, Gemma, CodeLlama, DeepSeek, and Qwen, available on Hugging Face Hub.
Is my data safe when using LM Pocket?
Yes, LM Pocket is designed for complete privacy. Every conversation stays on your device, with no servers, API keys, or accounts required. Your prompts and responses never leave your device.
How often is LM Pocket updated?
The latest version, 1.0.5, was released on August 8, 2026, indicating recent development activity. Updates are provided by the developer, Teo Bouancheau.
What devices is LM Pocket available on?
LM Pocket is currently supported on iPhone devices.
Can I connect other apps to LM Pocket?
Yes, LM Pocket includes a local API server that runs an OpenAI-compatible HTTP server on your device, allowing other apps that support the OpenAI API format to connect.
What are the system requirements for running models in LM Pocket?
Model performance in LM Pocket depends on your device's RAM and the size of the model. The app also features GPU acceleration for faster inference.
What is the age rating for LM Pocket?
LM Pocket has an age rating of 4+, making it suitable for a wide range of users.
