Enclave

Run LLMs locally in your browser, on your own GPU. Ask about any page — nothing ever leaves your machine.

Enclave lets you ask questions about any web page — using a large language model that runs entirely on your own device. No servers, no accounts, no data leaving your machine. WHAT IT DOES • Open the Enclave side panel and ask about the page you're reading • Select any text and right-click to explain or summarize it • Extract key information from long articles, docs, or research pages • Handles large pages via chunking and retrieval, so it works even on lengthy content RUNS 100% LOCALLY Enclave uses WebGPU to run open models (Llama 3.2, Gemma 2, Qwen3, and others) directly in your browser. Choose a model that fits your machine — from a lightweight 1B model to an 8B model for higher-quality answers. The only network request Enclave ever makes is the one-time download of the model weights you choose; after that, everything is cached locally and processed entirely on-device. PRIVATE BY DESIGN • No telemetry, no analytics, no accounts • Your page content, selections, and questions never leave your machine • Full details in the linked privacy policy REQUIREMENTS A Chromium browser with WebGPU support (recent Chrome, Edge, Brave, Arc) and a reasonably capable GPU — more VRAM lets you run larger, higher-quality models.