LLM Studio Go

Video studio and video generation (beta)
Turn an idea into an editable script, narration, pictures or clips, music, captions and a cover. Add your own media, keep projects and export a vertical MP4. FastWan 2.1 needs 8 GB RAM; FastWan 2.2 needs 12 GB. CPU video generation takes minutes per short clip and can heat the device. Download the required models first.

Meet your offline audio studio!
• Turn an idea into a song: local AI writes lyrics and creates music with vocals.
• Talk to your LLM offline, with live captions and word highlighting.
• Choose voices and languages.
• Explore music styles, vocal presets and longer tracks, or shape everything in PRO.
• Enjoy a full player with seeking, speed, repeat, saving and sharing.
• Faster ACE-Step song generation and audio model benchmarks.
Download models first; speed depends on your device.

An ad card may appear during audio generation. The ad-removal purchase also hides this placement.

Your ideas deserve a workspace that goes with you. LLM Studio Go brings local AI chat, documents, images and code to your Android device.

CHOOSE YOUR AI
Download quantized models from the built-in library or open a compatible GGUF file from your storage. Switch between models for different tasks, adjust generation settings and explore reasoning with supported models.

MAKE MORE OF YOUR DOCUMENTS
Attach PDFs, text documents and scans. Extract text with offline OCR, then ask your model to summarize, explain or find the details you need. English, Russian and Ukrainian OCR languages download on first use.

CREATE AND EXPLORE IMAGES
Turn a prompt into an image with supported local diffusion models. Save or share the result. With a compatible vision model, attach a picture and ask questions about what it shows.

FROM AN IDEA TO WORKING CODE
Read formatted answers with Markdown, tables and formulas. Write and run Python or JavaScript, and preview HTML/CSS/JavaScript projects in the app.

GO ONLINE WHEN YOU CHOOSE
Optional web search and webpage reading bring outside context into a conversation, with source links. Web access is off by default and can be controlled in settings. Online requests use external services.

KEEP YOUR WORK TOGETHER
Find conversations with search, organize them in folders and pin the ones you return to. Explore model benchmarks, device diagnostics and an optional authenticated local API. The interface supports 12 languages.

BUILT FOR LOCAL INFERENCE
No LLM Studio Go account or cloud inference subscription is required. Models and chat history are stored on your device. Local chat and image generation work offline after the required model files are downloaded.

Model files are downloaded separately and can be several gigabytes. Start with a smaller quantized model that fits your available RAM. Speed, supported features and image quality depend on the model and your device; large models may not fit every phone.

The model library contains ads, with an optional purchase to remove them. Advertising and purchase services use the network. Chats are not sent to advertising services. Optional web access and reports you choose to submit are covered in the Privacy Policy.

AI answers can be inaccurate. Check important information and use the report action to flag problematic responses.