Honey LLM - Offline AI

Ron.Bee

In-app purchases

About this app

Private AI that runs on your phone

Honey LLM brings useful AI directly to your Android device. Your conversations and supported documents are processed locally instead of being sent to a cloud AI service.

Download a local AI model once, then chat without an active internet connection.

No Honey account required.

Honey picks an AI for your phone

Local AI does not have to be complicated.

Choose Beginner mode and Honey evaluates your device and recommends a compatible AI model for you. If the model is already installed, you can start chatting right away.

Want more control? Switch to Advanced mode to choose from multiple local AI models and customize response behavior yourself.

Ask questions about your documents

Bring your own readable documents into Honey and ask what you actually want to know.

Honey supports:

PDF
DOCX
TXT
Markdown

You can:

Summarize a document
Ask freeform questions
Find important details
Ask about deadlines, payments, or timelines
Continue with follow-up questions
View supporting page or section references

Document processing stays on your device. Honey is designed to tell you when requested information is not available in the selected document sources rather than inventing an answer.

Image-only scans may require OCR before Honey can read their text.

Private chat without cloud inference

Once your AI model is installed, Honey can generate responses locally on your phone.

That means:

Your prompts are not sent to a cloud AI service
Your documents are processed locally
AI generation can work without an active internet connection
No Honey account or AI subscription is required

Internet access is still needed for activities such as downloading models, Google Play purchases, restoring purchases, and app updates.

Encrypted Private Vault

Honey also includes a local encrypted Vault for files you want to keep protected.

Vault files are stored in app-private encrypted storage using AES-256-GCM encryption backed by Android Keystore.

You can:

Import files into encrypted storage
Verify file integrity
Restore or export files when needed
Delete files from the Vault
Ask Honey questions about supported readable Vault documents

Your Vault remains local to your device.

Beginner when you want simplicity. Advanced when you want control.

Beginner mode handles model selection for you.

Advanced mode lets you explore Honey's local model catalog and control features such as:

AI model
Response style
Response length
Temperature
Top P

Downloaded models remain on your device for local use.

Try Honey before unlocking

Honey includes a free local AI demo so you can try real on-device inference before purchasing.

Unlimited use is available through a one-time in-app purchase.

No recurring Honey subscription.

Your local purchase price is displayed by Google Play.

Built for private, local AI

Honey is designed for people who want useful AI without automatically sending conversations and documents to a remote AI service.

Whether you want a simple AI assistant, private document questions, encrypted local files, or deeper control over local models, Honey gives you both an easy starting point and room to explore.

Your phone. Your AI. Your documents.

Ratings and reviews

larger AI models after download are slower but better quality responses. on my Google xl I use the 1.5 B model for best balance of speed and quality responses. the smallest one is fast but not as good quality wise as the bigger model

Tested on two phones S22 and M02. works great on newer phone, slower on cheaper devie. my S22 is my main phone and works perfectly! Fast, reliable, and private. definitely what the world needs right now

don't waste your time and data, you have to download a file more than 2GB larg inside this app, but that will not going to work at all, all it gonna do is crush

What’s new

New Beginner experience for faster, simpler setup
Smarter device-aware AI recommendations based on your phone
Improved on-device performance and efficiency
Better document summaries
Improved document Q&A and grounding
More reliable handling of whole-document questions
Improved responses when requested information is not present in a document
Refined Beginner model setup UI
Improved runtime configuration for supported devices
Additional stability, compatibility, and performance improvements