how to run gpt-oss locally — OpenAI's own open model, on your own machine

OpenAI released open-weight models you can run yourself. The 20B fits a 16GB card thanks to clever quantization; the 120B needs a serious rig.… Aliteq Lena Fischer · AI & Local Compute Editor

One command: ollama run gpt-oss:20b (OpenAI partnered with Ollama to distribute these).

gpt-oss-20b needs ~16GB VRAM — its MoE design + MXFP4 quantization pack it to ~12-13GB on disk.

gpt-oss-120b needs 60-80GB — a workstation/multi-GPU job, not a typical desktop.

Built for reasoning + agentic tasks — and works through a local OpenAI-compatible API.

Short on VRAM? CPU offloading works but runs slower.

Free, private, offline — OpenAI's models, running on your machine.

Read the full story

Aliteq how to run gpt-oss locally — OpenAI's own open model, on your own machine Read the full story on Aliteq