oobabooga/TextGen
Open-source desktop app for local LLMs. Text, vision, tool-calling, OpenAI/Anthropic-compatible API. 100% private.
TextGen, formerly text-generation-webui, is a free, open-source desktop app by oobabooga for running local LLMs privately on your own machine. It offers chat, an OpenAI- and Anthropic-compatible API, vision, tool-calling, and LoRA training across backends like llama.cpp and Transformers.
Our Review
With more than 47,000 GitHub stars, TextGen (formerly text-generation-webui) is one of the longest-running and most capable interfaces for local LLMs, and it ships frequent releases (v4.9 in May 2026). It runs fully offline with zero telemetry, which is the whole point for people who want models on their own hardware.
What TextGen does:
- •One-minute setup download a portable build for Windows, macOS, or Linux, unzip, and double-click; all dependencies are bundled.
- •Multiple backends llama.cpp, ik_llama.cpp, Transformers, ExLlamaV3, and TensorRT-LLM, switchable without restarting.
- •OpenAI and Anthropic-compatible API use it as a local drop-in replacement for those APIs, with tool-calling support.
- •Chat, instruct, and notebook modes talk to characters, follow instructions like ChatGPT, or free-form generate, with message editing and branching.
- •Vision and file attachments attach images for multimodal models and upload PDF or docx files to discuss their contents.
- •Tool-calling and MCP models can call custom Python tools including web search and page fetching, and MCP servers are supported.
TextGen ecosystem:
- •GGUF models from Hugging Face drop a GGUF file into the models folder and the UI detects it automatically.
- •Training and image generation fine-tune LoRAs and generate images with diffusers models from dedicated tabs.
- •Extensions community and built-in extensions add TTS, voice input, and translation.
Getting started:
Download a portable build from the releases page (CUDA, Vulkan, ROCm, or CPU-only), unzip, and run textgen. Download a GGUF model from Hugging Face into the user_data/models folder and the app picks it up. Hardware backends and training are documented in the wiki.
Limitations:
TextGen is licensed under AGPL-3.0, a strong copyleft license that matters if you plan to build a network service on top of it. Running larger models well needs significant RAM or VRAM, and managing GGUF quantization levels has a learning curve. With 800+ open issues and a fast release pace, the breadth of backends means occasional setup friction on specific hardware. It is a power-user tool rather than the absolute simplest local-LLM app.
Our Verdict
TextGen is the most capable open-source interface for running local LLMs in 2026, and its multi-backend design plus OpenAI-compatible API make it as useful as a server as it is as a desktop chat app.
For Developers, the local OpenAI- and Anthropic-compatible API is the headline: point existing code at TextGen instead of a paid endpoint and run everything privately, with tool-calling and MCP support intact. The choice of backends lets you tune speed and quality for your exact hardware.
For Vibe Builders, the one-minute portable install and GGUF auto-detection mean you can be chatting with a local model in minutes without a Python environment. The chat, vision, and file-attachment features cover most everyday private-AI needs.
Skip it if you want the absolute simplest one-click experience with no settings to learn, where a more opinionated app like LM Studio or Ollama with a UI may suit you better. The AGPL license is also worth checking if you plan to embed it in a commercial network service.
Frequently Asked Questions
Is TextGen (text-generation-webui) free?
Yes, TextGen, formerly text-generation-webui, is free and open-source under the AGPL-3.0 license, and it has more than 47,000 GitHub stars as of 2026. You can download and run it at no cost. Because it runs models locally on your own machine, you also avoid the per-token fees of cloud AI APIs.
What can TextGen do?
TextGen runs local large language models with chat, instruct, and notebook modes, plus vision for images and uploads of PDF and docx files. It exposes an OpenAI- and Anthropic-compatible API with tool-calling and MCP support, and it can fine-tune LoRAs and generate images. It supports multiple backends including llama.cpp and Transformers.
Who makes TextGen?
TextGen is created and maintained by the developer known as oobabooga, who launched it as text-generation-webui in late 2022. It has since become one of the most popular open-source tools for running local LLMs, with an active release cadence and a large community of contributors and extension authors.
What is the difference between TextGen and Ollama?
Ollama is a lightweight runtime focused on simple local model serving, while TextGen is a fuller desktop app with a rich UI, multiple backends, training, image generation, and extensions. Choose TextGen when you want maximum features and control in one interface. Choose Ollama when you want the simplest model-serving backend.
Do I need a GPU to run TextGen?
No, TextGen offers CPU-only portable builds, so you can run smaller models without a GPU, though a GPU with enough VRAM makes larger models much faster. It supports CUDA, Vulkan, and ROCm builds for NVIDIA, general, and AMD hardware. Models in the efficient GGUF format run well on modest machines.
What is TextGen?
TextGen, formerly text-generation-webui, is a free, open-source desktop app by oobabooga for running local LLMs privately on your own machine. It offers chat, an OpenAI- and Anthropic-compatible API, vision, tool-calling, and LoRA training across backends like llama.cpp and Transformers.
How do I install TextGen?
Visit the GitHub repository at https://github.com/oobabooga/textgen for installation instructions.
What license does TextGen use?
TextGen uses the AGPL-3.0 license.
What are alternatives to TextGen?
Explore related tools and alternatives on My AI Guide.
Open source & community-verified
AGPL-3.0 licensed: free to use in any project, no strings attached. 47,283 developers have starred this, meaning the community has reviewed and trusted it.
Reviewed by My AI Guide for relevance, quality, and active maintenance before listing.
Topics