Weave

Weave is a desktop application for working with artificial intelligence. It brings chat with language models, image generation and model management together in one place. It picks a suitable model for each request and creates images through ComfyUI.

What it does

• Chat with cloud and local models: the Mistral service, or built-in inference through llama.cpp right on your graphics card (Vulkan, CUDA, Metal on a Mac).
• Smart routing: the app decides which model suits a given message.
• Image generation through ComfyUI (SDXL, Flux, PuLID), including reference images, up to 40 at once.
• Context management: a gauge showing how full the model window is, and compaction so long conversations do not overflow.
• Automatic conversation titles and full-text search across the history.
• ComfyUI installation in a few clicks. The Windows installer picks CUDA or Vulkan itself based on the graphics card it finds.
• Updates are offered right inside the app.

Technology

The app is built on Tauri 2 (Rust and Svelte 5), stores conversations in SQLite and access keys in the system credential store. The code follows clean architecture.