Skip to main content
PromptQuorumBuilt for humans. Structured for AI.
Home/Power Local LLM
Mobile & Edge LLMs

Mobile LM Server Review: Android App for Local Models and a Phone-Hosted API

·6 min readBy Hans Kuepper · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI

Mobile LM Server is a closed-source Android app that says it runs GGUF and LiteRT-LM models offline and turns the phone into an OpenAI-compatible and Ollama-compatible API server. It is free to install but contains ads and in-app purchases, and its privacy policy lists ad-related data collection alongside the "fully local" pitch.

Key Takeaways

  • Developer: a developer account named "chaterminal" with an address in China on the listing; the developer website returned an error when checked.
  • Reach: Google Play shows 1K+ downloads and a 3.8 average from 45 ratings, so the public track record is small.
  • Activity: version 1.1.10 was updated on 22 October 2026, so the app is under active development.
  • Platform: Android 7.0 or later through Google Play only; no iPhone, desktop or web build was found.

📍 In One Sentence

Mobile LM Server is a closed-source Android app that, according to its developer, runs open-weight models on the phone with llama.cpp or LiteRT-LM and offers OpenAI-style and Ollama-style API endpoints to other devices.

💬 In Plain Terms

You install it from Google Play, load a model and either chat in the app or point a desktop tool at the phone. Hybrid cloud routing and online search are optional extras, and ads are part of the free version.

📌Note: This review is based on the Google Play listing and the privacy policy it links, checked on 26 October 2026. PromptQuorum did not install the app, run a model, call the API or measure speed, battery use or network traffic.

Read in:
🇺🇸en🇩🇪de🇫🇷fr🇯🇵ja🇨🇳zh🇪🇸es🇧🇷pt🇸🇦ar🇰🇷ko

Mobile LM Server (Google Play package com.chaterminal.mobilellm, developer account "chaterminal") is an Android app that, according to its listing, runs GGUF and LiteRT-LM models on the phone and exposes them through OpenAI-compatible and Ollama-compatible endpoints. It contains ads and in-app purchases. This review is based on the Google Play listing and the privacy policy it links, as checked on 26 October 2026; it involved no hands-on use.

Mobile LM Server Review: Android App for Local Models and a Phone-Hosted API

What Is Mobile LM Server?

Mobile LM Server is an Android app that, according to its Google Play listing, runs open-weight language models directly on the phone and can act as a small inference server for other devices. The listing names two engines, LiteRT-LM and llama.cpp (GGUF files), and lists Llama, Gemma, Phi, Mistral, Qwen and DeepSeek distilled models as supported families. It also lists a built-in chat, prompt compression, device monitoring (GPU, memory, speed, temperature) and optional "hybrid AI routing" that can send long queries to cloud models.

The Play title is "Mobile LM Server - Local AI" and the description calls it "Mobile LLM Server"; the developer account is "chaterminal". It is unrelated to the open-source GitHub project named chaterminal (a terminal chat tool) and is a different app from OlliteRT, which also serves a phone-hosted API.

Get It

Mobile LM Server is installed from Google Play like any other Android app; no sideloading or terminal is involved. The developer's website link on the listing returned an access-denied error on 26 October 2026, so the privacy policy page is linked separately below.

Platform
Where to get it
AndroidGoogle Play (free install with ads and in-app purchases, Android 7.0 or later)
Websitechaterminal.com (linked from the listing; returned an access-denied error on 26 October 2026)
Privacy policyPrivacy policy (page effective April 2026)

Version reviewed: 1.1.10, shown on Google Play as updated on 22 October 2026. The only release note shown says it fixed online search in the built-in chat for GGUF models.

Features Confirmed by the Sources

Every item below comes from the Google Play description or its release note; none was tested.

  • Local inference. Models run on the device through LiteRT-LM or llama.cpp; the description says no internet connection, login or data sharing is needed for local inference.
  • API server. The phone can expose OpenAI-style and Ollama-style endpoints so desktop tools, bots and scripts can use it. The privacy policy describes this PC-to-phone traffic as local network (LAN) communication.
  • Built-in chat. A chat screen for the loaded model; the 1.1.10 release note adds online search to it for GGUF models.
  • Hybrid routing. When local context limits are reached, queries can optionally be offloaded to cloud models; the listing does not say which providers or how they are configured.
  • Prompt compression and monitoring. Long system prompts and tool descriptions are compressed before execution, and the app shows GPU use, memory, generation speed and temperature.

Privacy, Ads and Network Use

According to the privacy policy, prompts, model output and conversations are never uploaded to any server, but the app does collect usage data for ads. The policy (effective April 2026) lists app-launch, model-download and model-start events, crash logs and approximate, city-level location, used for ad serving, content optimization and service improvement.

The Google Play Data safety section reads differently: it says "No data collected" and also lists location among data types the app may share with third parties, notes that data is encrypted in transit and states that data cannot be deleted. The listing carries "Contains ads" and "In-app purchases" labels; the description does not say what the purchases unlock. One reviewer wrote in September 2026 that hosting the API required watching an ad, and the developer replied that ads cannot be removed for now and that the latest version can save models on the device. Both statements are the parties' own and were not tested here.

📌Note: The app is closed source, so none of these claims can be checked in code, and PromptQuorum has not audited network traffic or permissions. Anyone handling confidential prompts should verify behavior on their own device first.

Who Should Use It

  • Android developers who want to try a phone as a local API endpoint. The listing offers OpenAI-style and Ollama-style endpoints with two engines.
  • GGUF users who want a phone-side runner. llama.cpp support is named in the listing; whether your model file loads is untested.

What We Could Not Verify

  • Hands-on performance. Speed, battery use, heat and answer quality were not measured; nothing in this review comes from running the app.
  • The models. The listing names families but not specific models, sizes or download sources; one reviewer mentions large downloads from Hugging Face.
  • The API. Which endpoints work, whether authentication exists and how ad prompts affect the server were not tested.
  • Hybrid routing. Which cloud providers receive data when routing is on, and what is sent, is not described.
  • Hardware needs. No RAM or storage requirement is stated; the only stated floor is Android 7.0.
  • License and source code. No repository was found; the policy footer says "Open Source Project" but links none, so the app is treated as closed source.
  • Permissions. The permission list was not inspected.

Competitors and Alternatives

OlliteRT

Platforms:
Android
Price / license:
Free / Apache 2.0
Key difference:
Open-source phone-hosted OpenAI-compatible server built on LiteRT-LM
Articles about OlliteRT (2)

Competitor details change often; confirm each app's current price, license and platforms on its own listing.

Compare in directory

Frequently Asked Questions

Is "Chaterminal" the app or the developer?

Chaterminal is the developer account name on Google Play; the app is titled Mobile LM Server - Local AI and uses the package name com.chaterminal.mobilellm.

Is Mobile LM Server open source?

No repository was found. The privacy policy footer calls it an open-source project but links no code, so this review treats it as closed source until a repository is published.

Does it work on iPhone or desktop?

No such build was found; the listing is Android only.

Verdict

Mobile LM Server is an actively updated Android app that combines two inference engines, a built-in chat and an OpenAI-style API in one free install, but the offline claim and the API are untested here, and the free version carries ads and a privacy policy that lists ad-related data collection. It suits developers who want to experiment and accept a closed-source build from a small developer whose website returned an error. Anyone who wants an open-source server should compare OlliteRT; for plain chat, see PocketPal AI.

Sources

  • Mobile LM Server on Google Play — description, version, update date, release note, labels, Data safety section and reviews, checked 26 October 2026.
  • Mobile LM Server privacy policy — data processing, collected information and LAN communication; page effective April 2026.
  • chaterminal.com — the developer website linked from the listing; it returned an access-denied error on 26 October 2026.

← Back to Power Local LLM

MCP