OfflineGPT

OfflineGPT

Private, on-device AI—chat, image generation, voice, and tools for Android (Google Play).

Offline-first with optional cloud features (your API keys, optional web search). No ads. Billing via Google Play.

iPhone or iPad? Offline AI Studio — separate product page & App Store listing.

Built for Android

MNN image pipelines, Play Billing, and tool limits tuned for this build.

  • Android 8.0+ (API 26+), 64-bit ARM
  • Image generation: MNN diffusion (AbsoluteReality, Anything V5)
  • Google Play Billing for Pro (one-time)

What OfflineGPT can do

On-device first—with honest optional online modes

Offline-first AI

Core chat, tools, and voice run on-device after you download models. Optional API mode and web search only when you turn them on.

Privacy-minded

Local chats and generations stay on your phone by default. If you use cloud providers or web search, data goes to those services under your control—not through a CodeAlp chat backend for those flows.

Whisper voice input

Dictate prompts with on-device speech-to-text. Many languages; download size depends on the voice model you install.

On-device image generation

MNN-based diffusion (AbsoluteReality, Anything V5)—tuned for Android performance and battery.

AI tools & CodeLab

Summarize, tone, grammar, posts, CodeLab, and more. Free vs Pro varies by tool—see the table below.

Thinking models

Qwen3 thinking variants and SmolLM3 can show step-by-step reasoning when you pick a compatible model.

Themes & languages

9 free + 7 Pro themes. UI in multiple languages—see the in-app list.

Custom GGUF imports

Bring compatible GGUF chat models. Pro unlocks managing many imports; free keeps one catalog model plus one custom import (replace rules in Model Manager).

How it works

Four steps

1

Install from Google Play

OfflineGPT targets Android 8.0+ on 64-bit ARM (arm64-v8a).

2

Download models

Chat, image, and voice assets—sizes from under 1 GB to several GB per model.

3

Chat locally

Run prompts on-device. Enable API providers or web search only if you want cloud-backed features.

4

Go Pro if you like

One-time Play purchase unlocks full model slots, premium themes, longer context, and Pro-gated tools.

Why OfflineGPT

Most AI apps default to the cloud. OfflineGPT defaults to your device—and stays honest when you opt in to the network.

Your keys, your device, your choice of models.

AI tools (Android)

Free vs Pro follows the current app build. Document, Email, Brainstorm, and custom tools are Pro-gated on Android.

ToolAccess
SummarizeFree
ToneFree
GrammarFree
ExplainFree
Post / socialFree
CodeLabFree
Document & OCRPro
Email writerPro
BrainstormPro
Custom tools (create / manage)Pro

On-device models (catalog)

Approximate download sizes. Rocket-3B is Android-only in this lineup.

Chat models

LFM2.5 1.2B

LiquidAI

~731 MB

Llama-3.2-1B

Meta

~1.35 GB

Gemma-2B

Google

~1.85 GB

Qwen2.5-3B

Alibaba

~2.1 GB

Rocket-3B

Community

~3.1 GB

Thinking models

Qwen3 0.6B Thinking

Alibaba

~640 MB

SmolLM3-3B

Hugging Face / ggml-org

~1.9 GB

Qwen3 1.7B Thinking

Alibaba

~1.8 GB

Qwen3 4B Thinking

Alibaba

~2.6 GB

Image generation (MNN)

AbsoluteReality

Photorealistic (MNN)

~1.25 GB

Anything V5

Anime-style (MNN)

~1.25 GB

Voice

Whisper (voice input)

OpenAI Whisper weights

~80 MB (typical base bundle)

Pro: import additional GGUF chat models from compatible sources (subject to RAM and validation).

Free vs Pro (Android)

Current Google Play build

FeatureFree Pro
On-device text/chat models1 catalog model (replace to switch) + 1 custom GGUF importInstall and switch between many models from the manager
On-device image models at once1 (replace to switch)Unlimited installed
Local chat context (typical max)Up to 2,048 tokensUp to 32,768 tokens (local); API limits follow provider
System prompt editingPresets visible; editing ProFull edit + custom saved prompts
User prompt templatesMax. 5 customUnlimited
Premium themes7 Pro-only themes lockedEvery theme
Cloud API providers & web searchPro onlyAdd keys for OpenAI, Gemini, Groq, etc.; optional web search when enabled
AdsNoneNone
Purchase typeOne-time Pro (no subscription)

One-time Pro purchase on Google Play—price shown in the store.

Privacy policy

Frequently asked questions

Do I need the internet?

You need a connection to download models and app updates. After that, core on-device features work offline. Optional API mode and web search need internet when you use them.

Is OfflineGPT free?

Yes. Free tier includes on-device use with limited model slots and tool locks—see the tables. Pro is a one-time Google Play purchase.

Is there an iPhone or iPad version?

Yes. On the App Store the app is listed as Offline AI Studio—the same product philosophy, built for iOS with Core ML. See the dedicated page. Offline AI Studio product page.

Which chat models are available?

GGUF catalog (LFM2.5, Llama 3.2, Gemma 2B, Qwen2.5-3B, Rocket-3B, Qwen3 thinking family, SmolLM3). Exact filenames match the in-app downloader.

How does voice input work?

Download Whisper assets, grant microphone access, then dictate. Audio is processed for transcription; default stays on-device unless you route through a cloud provider.

What are the AI tools?

Shortcut flows for summarize, grammar, tone, posts, code help, documents, and more. Document, Email, Brainstorm, and custom tool authoring follow the Free/Pro table.

How does privacy work?

Default local use keeps prompts and history on your device. Full details:codealp.ch/privacy-offlinegpt.

What devices are supported?

Android 8.0+ on 64-bit ARM (arm64-v8a). Pick models that fit your RAM.