MLC Chat

Name: MLC Chat
Rating: 4.2 (84 reviews)
Author: Tianqi Chen

By: Tianqi Chen App

Chat with Open Language Models

Features & Capabilities

MLC Chat lets you chat with open language models locally on iPhone and iPad. After you download a model, all processing runs on-device, with no server support and no internet connection needed, ensuring privacy and that data never leaves your device.

As part of the open-source MLC LLM project, it supports multiple open-model architectures on diverse hardware backends and native applications. The app runs different open models on your phone, and performance depends on your device's VRAM. It is intended for non-commercial use and respects model licenses as applicable.

User Growth & Download Statistics

App AI Open Source Privacy & Security Developer Tools Programming

By:

Tianqi Chen

Rating:

4.20

(84)

Version:

1.6 Last updated: 2024-09-27

Version code:

869309693

Creation date:

2023-05-31

Compatible devices:

Size:

1.71GB

URLs:

Website ,Privacy policy

Full description:

Source:

Apple Apps Store

Data ingested on:

2026-07-15

Compare stats and ranking:

Ranking

# 166,830 ▲ 2,872

For Developers

Monetize with App Pass Boost with Extension Booster

Contact the developer

Chrome-Stats does not own this Apple app. Please use these information below to contact the Apple app developer.

Developed by:

Tianqi Chen

Apple Apps Store

https://apps.apple.com/us/app/mlc-chat/id6448482937

Website:

https://mlc.ai/mlc-llm/

User Reviews

Users praise MLC Chat for on-device, offline inference with no subscription and good on-device speed across several models. However, reliability and usability suffer from crashes, heavy memory/VRAM demands, and unclear model-import options. Conversations aren’t saved, reducing long-term usefulness. Many requests include history/shortcuts support and an OpenAI-like API. Overall, it’s an impressive proof-of-concept for local LLMs, but it needs better reliability, model-management, and feature completeness to appeal to broader users.

Pros

Completely offline/local inference with no subscription
Supports multiple models (Llama, Gemma, Phi, Mistral, etc.)
Surprisingly fast responses on iPhone; strong on-device performance
Free to use; no cloud dependency or subscription
Demonstrates on-device LLAMA2 viability on iPhone

Cons

No conversation history saving; chats aren’t persisted
Difficult to import/add new/custom models; unclear instructions; removal of custom model downloads
Stability issues: crashes with several models on various devices due to memory/VRAM constraints
High VRAM/memory requirements limit device compatibility and can cause throttling/heating
Missing features requested by users: shortcuts, per-model settings, and OpenAI-like API integration

Recent reviews

Overly censored and a waste of time. I get done censorship, but when you over censor everything, you become useless!!! Try putting rules in place to block the underage and allow the adults to be adults and make their own decisions. If kids gain access, that’s the parent’s fault! We shouldn’t have to suffer because people don’t know how to correct their bad behaved children!

by Bl*****, 2026-06-09

Look it works so I’ll give it that but you can only use one of 5 models. Maybe it’s hard to code the ability to import others but add that and it’s 5 stars