gpt-oss-20b

Run gpt-oss-20b as a local LLM on your device. llama.cpp runs directly inside your browser, using WebGPU or CPU where supported. No remote inference server is required.

Hugging Face repository: ggml-org/gpt-oss-20b-GGUF.

JavaScript is required to run this model. Review the model and its size, then choose Run in your browser to prepare it and start a chat. If the app does not load, reload this page or open Naidan.