Run Qwen, Gemma & Gemini Nano offline. Chat with your documents. 100% Private
PocketGem AI Agent turns your phone into a private, on-device AI workstation, running Qwen 2.5, Gemma, and Gemini Nano locally with zero data tracking. It enables chat with documents via local RAG and supports multiple models and builtβin toolsβall without internet access. With 1,000+ total downloads and 460 in the last 30 days, it signals steady demand for offline, privacyβfirst AI.
Pros & Cons
Fully offline on-device AI runs without internet, ensuring privacy.
Multiple models supported on-device (Qwen 2.5, Gemma, Gemini Nano) with Hugging Face integration.
Local document chat (RAG) with citations and private data handling.
Encrypted vault via Android Keystore protects API keys and data.
RAM and storage requirements are relatively high (minimum 6GB RAM; 4β8GB free storage) which may limit on-device compatibility.
Current rating is 0.0 with 0 ratings, offering limited social proof.
On-device operation may lack some cloud-based feature breadth and collaboration options found in cloud-enabled rivals.
About PocketGem AI Agent
PocketGem AI Agent is a productivity app developed by VaneSpark Apps.
Download Statistics
PocketGem AI Agent has been downloaded 2.2 thousand times. In the last 30 days, the app was downloaded 460 times.
User Ratings
The app has no ratings yet.
App Information
PocketGem AI Agent is FREE to download. The APK download size is 68.64 MB. The latest version available is 1.0.27. The last update was on April 6, 2026.
Technical Requirements
The app has a content rating of Everyone. The app has been available on Google Play November 2025.
Description
PocketGem transforms your phone into a powerful, private AI workstation. Run the latest Large Language Models (LLMs) like Qwen 2.5, Gemma, and Gemini Nano locally on your device.
Enjoy a completely offline AI agent with zero data tracking, zero latency, and zero subscription fees. Whether you need a secure coding assistant or a private document analyzer, PocketGem runs entirely on your hardware.
π TOP FEATURES
π 100% Offline & Private • Local Inference: AI models run directly on your CPU, GPU, and NPU. • No Internet Needed: Chat, analyze docs, and run tools without Wi-Fi. • Zero Tracking: No telemetry, no cloud servers, no analytics. • Encrypted Vault: API keys and data are secured via Android Keystore.
π Chat with your Documents (RAG) • Interactive Analysis: Import files and chat with your documents instantly. • Local RAG: Turn PDFs, DOCX, and TXT files into a searchable Knowledge Base. • Instant Answers: Ask questions about your files and get citations for every answer. • Privacy First: Your sensitive documents never leave your device.
π€ Multi-Model Support (New!) • Qwen 2.5: Run the latest high-performance models instantly—no account required. • Gemini Nano: Experience Google's efficient on-device AI. • Gemma 3 & 3n: Optimized for mobile with INT4/INT8 quantization. • Hugging Face: Download additional models directly from the hub.
β‘ AI Agent & MCP Tools • Autonomous Agent: Complex problem solving with "Thought Chains." • MCP Support: Connect to external data via Model Context Protocol. • Built-in Tools: Calculator, Weather, and private Brave Search.
βοΈ Power User Controls • Custom Context: Up to 32K context window for long documents. • Sampler Settings: Fine-tune Temperature, Top-P, and Top-K. • Tech Stack: Switch between CPU, GPU, and NNAPI acceleration.
π± REQUIREMENTS • RAM: 6GB (Minimum), 8GB+ (Recommended for Qwen/Gemma 3n), 12GB+ (Best Performance). • Storage: 4–8GB free space for model weights. • Android: 8.0+
π― WHY POCKETGEM?
Most AI apps send your data to the cloud. PocketGem keeps your intelligence local. Perfect for privacy advocates, researchers, and anyone who wants a "brain in their pocket" that works anywhere, anytime.
Download now to run Qwen 2.5 and Gemma offline!
Recent changes: 1.0.27: Gemma 4 E2B with vision added 1.0.26: Added Vision Support For Gemma 3n! 1.0.25: Added support for using your own LiteRT models 1.0.24: Easily managing settings with profile management 1.0.23: Update app icon 1.0.22: RAG improvements and crash fixes 1.0.21: Clarify classic RAG on Agent screen 1.0.20: More models added (no Hugging Face account required)!
Get a detailed PDF report for PocketGem AI Agent with download trends, rating history,
and key performance statistics β useful for competitive research or tracking your own app.
Learn more
Are you the developer of this app? Join us for free to see more information about your app and learn how we can help you promote and earn money with your app.
Love it! Easy to use LLM on device app. I can use AI with my documents
β β β β β Francis OlszanskiΒ· Jan 18, 2026
Fantastic app guys! I'm using the largest Gemma 3 4b vision model on my Pixel 8Pro, and it runs very well using the NPU acceleration It would be great if you could add TTS integration π Keep up the good work!
β β βββAbhishek PatelΒ· Dec 21, 2025
How to download llm model not showing any option to download the models
Google Play Rankings for PocketGem AI Agent
This app is not ranked
Technologies used by PocketGem AI Agent
PocketGem AI Agent is requesting 10 permissions and we don't have library information (yet).
Show details
Trust & Safety: Privacy-focused design with offline operation and zero telemetry per the description. Requests are aligned with offline AI needs (network state, internet, wake lock, foreground service, boot broadcast, etc.). Data and API keys are secured via an encrypted vault using Android Keystore.
Each subscription will automatically renew 3 days before the expiration date for
the same time period. Subscriptions can be cancelled at any time before the renewal.