Loading…
CompareOpen in Google Play
This app isn't in any of our charts yet.
Fixed Pocket TTS crashing Added quantized version of Supertonic TTS model Bumped llama cpp
Run local AI, generate images and music, translate, transcribe, vibecode offline
LLM Hub brings local AI to your device with private, on-device tools for LLM chat, image and music creation, translation, writing and speech. Choose the models you want, then use core offline AI features without sending prompts to a cloud inference service. Run a local LLM for everyday questions, creative work and productivity while keeping control of your data. Local LLM chat • Chat with local models including Gemma and Llama • Import compatible GGUF models or discover models from Hugging Face • Continue multi-turn conversations with RAG memory • Ask about text and images, use optional web search and hear replies with local text-to-speech AI Agent and MCP • Let an on-device AI Agent use maps and built-in device tools • Connect compatible Model Context Protocol (MCP) servers • Review and approve every MCP tool call before it runs • Generate, inspect and edit Termux commands before execution Create on your device • Offline image generator: turn text prompts into original images • Music generator: make original music and sound effects locally • Image upscale: enhance images up to 4× with AI super-resolution • creAItor: design custom AI personas with specialized instructions • Vibe Coder: describe an app idea and preview generated HTML and JavaScript Language, writing and voice • AI Translator: translate text, image text with OCR, and audio across 50+ languages • AI Writing Assist: summarize, expand, rewrite, improve grammar and generate code • Whisper Transcriber: use on-device Whisper transcriber models to convert recordings and audio files to text • VibeVoice: use hands-free AI voice chat • Scam Detector: check suspicious messages and images for phishing risks Private by design • Local AI and offline LLM inference • No account required • Conversations and generated content stay on your device during local inference • Open-source and transparent Internet access is required to download models and use optional online features such as web search or remote MCP servers. Feature availability, acceleration and speed depend on your device, chipset and selected model. Some features require Premium.