Guides

On-device Models

Run language models entirely on your phone with LiteRT-LM, including offline use and background downloads.

On-device models let Mobile Agent run a language model directly on your phone's hardware with the LiteRT-LM engine. No internet connection, no remote servers, and no tokens or messages leaving your device.

How it works

Models are packaged as .litertlm files and served from the Mobile Agent on-device model catalog (on-device-models.json), which the app fetches from the repository and caches in memory for 30 minutes, falling back to the bundled copy when offline. Downloads run through a native Android WorkManager worker, so a large model can download in the background while you keep using the app.

Model files are cryptographically verified: every entry pins an exact sha256 hash and sizeBytes, so a corrupted or tampered download fails safely instead of loading.

Requirements

Each model lists a minimum RAM requirement (minRamBytes). Some are based on the model publisher's published minimum, others are derived from benchmarks or comparable models. Before enabling a model, Mobile Agent checks your device against this figure — a model that won't run well is not offered.

Recommended guide (approximate):

RAMComfortable models
6 GBQwen3 0.6B, LFM 2.5 1.2B, Qwen2.5 1.5B, Phi-4 Mini, DeepSeek R1 1.5B
8 GBQwen3 4B, SmolLM3 3B, Gemini 4 E2B, Qwen3 1.7B, DeepSeek R1 1.5B
12 GBGemma 4 E4B, Qwen3 8B, DeepSeek R1 7B
16 GBGemma 4 12B

Larger models also report a low-memory context window that the app falls back to when available memory is tight, so you keep functional tool use even under pressure.

Current catalog

ModelParamsQuantizationContextToolsReasoningMin RAM
Gemma 4 E2B2.3Bmixed 2/4/8-bit32KYesNo8 GB
Gemma 4 E4B4.5Bmixed 4/8-bit32KYesNo12 GB
Gemma 4 12B12Bmixed 4/8-bit32KYesNo16 GB
Qwen3 0.6B0.6Bint42KNoNo6 GB
Qwen3 1.7B1.7Bint84KNoNo8 GB
Qwen3 4B4Bint42KNoNo8 GB
Qwen3 8B8Bint42KYesYes12 GB
Qwen2.5 1.5B Instruct1.5Bint84KNoNo6 GB
Phi-4 Mini3.8Bint84KNoNo6 GB
DeepSeek R1 Distill 1.5B1.5Bint84KNoYes6 GB
DeepSeek R1 Distill 7B7Bint4 block-324KNoYes12 GB
SmolLM3 3B3Bint4 block-324KNoNo8 GB
LFM 2.5 1.2B Thinking1.2Bint44KNoYes6 GB

Note: The catalog changes over time as models are added. Two engines are used: CPU for the smallest models (Qwen3 0.6B, LFM 2.5) and the GPU/LiteRT runtime for the rest. The capabilities.tools and capabilities.reasoning flags are shown in the UI so you know before you download whether a model can call tools or reason.

Install and use

  1. Open Settings → Providers → On-device.
  2. Tap a model card. Review its size, context window, minimum RAM, license, and capabilities.
  3. Download the model to your device. The worker runs in the background — you can leave the screen.
  4. When the download finishes, select the model in the composer's model picker.

Going fully offline

Because the runtime, tool logic, and model all live on-device, you can:

  • Disable Wi-Fi and cellular data and keep chatting with an on-device model.
  • Let scheduled jobs run offline (though jobs that call cloud MCP servers will report connection failures).
  • Keep your conversations, memory, and workspace files private on the device.

Licensing

Each model carries its own weight license — Gemma, Apache-2.0, MIT, or LFM Open. The license is shown on the model card before download. Check the individual model pages on Hugging Face for the exact terms that apply to your use case.