Text Generation WebUI

Formerly called text-generation-webui and now renamed TextGen, oobabooga's project is a local workbench for chatting with open-weight language models, offered as a portable desktop app or a browser UI on port 7860. Portable builds for Windows, Linux and macOS bundle everything needed to run GGUF files through llama.cpp on CUDA, Vulkan, ROCm or plain CPU; the full install adds Transformers, ExLlamaV3, ik_llama.cpp and TensorRT-LLM loaders, LoRA training, a diffusers image generation tab and community extensions for speech and translation. It accepts images for vision models, PDF and docx attachments, tool calls written as single Python files or served by MCP servers, and an optional web search tool. Its HTTP API imitates the OpenAI and Anthropic endpoints, so existing client code can point at your own machine. Hardware needs follow the model: a small quantized model runs on CPU, larger ones want a GPU with enough VRAM. Code is AGPL-3.0, free to use at work, but a modified version offered over a network must publish its source. Best for hobbyists and developers who want one private, telemetry-free front end for local models.