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

On Device LLM Review: Free Offline Android Chat App

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

On Device LLM is a free, closed-source Android app that says it runs open-weight models such as Llama 3.2 on the phone with no account. The project is brand new, so its claims rest on a listing and a policy that no one has independently tested.

Key Takeaways

  • Developer: an individual in India, listed on Google Play as Raushan kumar Pandey; the store page also shows the byline "Feather AI".
  • Age and reach: the first release is version 1.0, last updated on 11 September 2026, and Google Play shows 100+ downloads, so there is no track record yet.
  • Platform: Android 7.0 or later on 64-bit devices, through Google Play only; no iPhone, desktop or web build was found.
  • Maintenance: the last update is 45 days old as of 26 October 2026, so no stale warning applies.

📍 In One Sentence

On Device LLM is a closed-source Android app that, according to its developer, runs open-weight language models on the phone with no account, no ads and no server.

💬 In Plain Terms

You install it from Google Play, download one of seven models (or paste a link to your own GGUF file) and chat with it; the developer does not name the software that runs the models.

📌Note: This review is based on the Google Play listing and the developer's website and privacy policy, checked on 26 October 2026. No hands-on use, benchmarking or network inspection was done.

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

On Device LLM is a free Android chat app that, according to its Google Play listing and privacy policy, runs open-weight language models on the phone without an account. Google Play titles it "Local LLM"; it is a different app from the iPhone and Mac apps On Device AI and AI on Device. This review reflects the store listing and the developer's website and privacy policy, as checked on 26 October 2026.

On Device LLM Review: Free Offline Android Chat App

What Is On Device LLM?

On Device LLM is an Android chat app that, according to its Google Play listing, answers prompts with open-weight models running on the phone instead of a cloud service. The listing is titled "Local LLM", while the developer's website and privacy policy call it "local LLM" and describe on-device AI chat for Android; this review uses "On Device LLM" to tell it apart from the iPhone and Mac apps On Device AI and AI on Device, which have a different developer, platform and package name.

The listing names its developer as Raushan kumar Pandey in India; the Android package is com.ondevicellm. The app is published only on Google Play.

Get It

On Device LLM is installed from Google Play like any other Android app; no terminal or account is involved. The first model download is 369 MB or more, so the first run needs a connection and free storage.

Platform
Where to get it
AndroidGoogle Play (free install, 64-bit Android 7.0 or later)
Websiteon-device-llm-ff80d.web.app (developer site with legal pages only)
Privacy policyPrivacy policy (last updated 24 August 2026)

Version reviewed: 1.0, shown on Google Play as last updated on 11 September 2026; the release notes call it the first release. The listing names no source repository.

Features Confirmed by the Sources

Every item below comes from the Google Play description or the privacy policy.

  • Seven built-in models. SmolLM2 360M, Qwen2.5 0.5B, TinyLlama 1.1B, Llama 3.2 1B, Qwen2.5 1.5B, Gemma 2 2B and Llama 3.2 3B, each showing its download size before you commit; models can be switched without leaving the chat.
  • Your own models. You can paste a direct link to any GGUF file.
  • Saved chats. Conversations are stored in a database on the device, titles are written by the model, markdown and code blocks render with one-tap copy, and all chats can be cleared at once.
  • Optional local API. An OpenAI-compatible HTTP endpoint on port 11434 lets curl, the Python SDK or your own scripts use the phone's model over Wi-Fi or USB, with streaming. It is off by default and shows a notification while active.
  • Requirements. The listing states roughly 1 to 3 GB of free RAM depending on the model, and says larger models are noticeably slower on older phones.

Privacy and Network Use

According to the privacy policy, messages and replies stay in a private database on the device, and no prompt is ever sent to a server. There is no account, and the listing says there are no ads. The Google Play Data safety section says "No data collected" and "No data shared with third parties".

The policy is more detailed than that card. It says the app uses Google Firebase Analytics to record anonymous events: event names, the model name, message length in characters (not the text), durations, an anonymous app instance ID and standard device information. It also lists model downloads from the host you select (by default huggingface.co) and a read-only Google Cloud Firestore request at startup for the model catalogue and update checks. The two statements differ on analytics, so the policy is the fuller source.

The optional local API has no authentication: the policy says anything on your network that can reach the phone can send prompts and read replies, and advises enabling it only on networks you trust.

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

Who Should Use It

  • Android users who want a free offline chat app with their own choice of models. The GGUF link import is the main reason to try it.
  • Developers who want a phone-hosted endpoint on a trusted network. The local API is the other.

What We Could Not Verify

  • Hands-on performance. Speed, battery use and answer quality were not tested, and nobody here confirmed that chat works in airplane mode.
  • The runtime. The listing says models are GGUF files but does not name the inference engine.
  • License and source code. No public repository was found, so the app is treated as closed source; each model also carries its own license from its publisher.
  • In-app purchases. The install is free and the description names no purchases or ads, but the Data safety card does not state a business model.
  • The developer contact. The privacy policy shows a placeholder contact address (SUPPORT@EXAMPLE.COM); the support email is on the Google Play listing.

Competitors and Alternatives

Private Mind

Platforms:
iOS, Android
Price / license:
Free / MIT
Key difference:
Open-source offline chat with on-device document Q&A and benchmarks
Articles about Private Mind (2)

Also mentioned in:

+9 more not shown

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

Compare in directory

Frequently Asked Questions

Is On Device LLM the same as On Device AI or AI on Device?

No. Those are separate iPhone and Mac apps from other developers; this one is an Android app with the package name com.ondevicellm.

Can I use a model that is not in the list?

The listing says you can paste a direct link to any GGUF model. Whether a given model runs well depends on your phone's RAM, and its license is set by its publisher.

Does it work on iPhone or desktop?

No such build was found; the developer's website describes it as on-device AI chat for Android.

Verdict

On Device LLM is a free Android chat app with a clear offline pitch, a GGUF link import and a local API, but it is newly released, closed source and has 100+ downloads, and its offline and privacy claims are untested. It suits readers who want to try a small, account-less assistant and accept those risks; anyone who wants an open-source, established client should compare PocketPal AI or Private Mind.

Sources

← Back to Power Local LLM

MCP