MNN Chat

jindesong

About this app

This is a multimodal language model (LLM) Android app

Multimodal Support: Enables functionality across diverse tasks, including text-to-text, image-to-text, audio-to-text, and text-to-image generation (via diffusion models).

CPU Inference Optimization: MNN-LLM demonstrates exceptional performance in CPU benchmarking in Android.

Broad Model Compatibility: Supports multiple leading model providers, such as Qwen, Gemma, Llama (including TinyLlama and MobileLLM), Baichuan, Yi, DeepSeek, InternLM, Phi, ReaderLM, and Smolm.

Privacy First: Runs entirely on-device, ensuring complete data privacy with no information uploaded to external servers.

Ratings and reviews

Very fast. BUT buggy mess, specially with qwen models.I tried feeding the model with a prompt file with image path, but the model produces gibberish or repetitive letters. Only when I changed "memory": from low to normal the model finally worked. The downside is it became super slow. Please help. My device is poco x3 gt 8gb lpddr4, mali g77 mc9, dimensity 1100.

MNN Chat is a clean, fast, and surprisingly capable AI chat app. Responses are quick, the interface is simple, and it’s easy to jump in and get useful answers without any friction. Great for everyday questions, ideas, and quick problem-solving. 5/5 - simple, fast, and actually useful.

Great performance but can't search for non MNN models and severely lacking even basic features. In response to the dev: The non MNN thing is just a fact... and there are no options for KV Cache quant, Chat Truncation, Response/Reasoning Limits, Hide Reasoning, set Seed, edit/delete prompt, edit/delete/regenerate output, continue generating, or Context usage.

What’s new

Support Qwen3.5 ThinkSwitch, Fix some bugs