On-device AI models running entirely in your browser
Real-time webcam and video analysis with a vision-language model. Ask questions about what the camera sees.
Open demo → OCR · Qwen3.5Extract text from images and PDF documents on-device, with no upload to any server.
Open demo → Vision · LFM2.5-VLLiquidAI's 450M vision-language model describing your webcam or video in real time, streaming token-by-token from the GPU.
Open demo → Chat · Qwen3.5Multi-turn text chat with Qwen3.5, with optional step-by-step reasoning, streaming token-by-token from the GPU.
Open demo →