Skip to content

gpt-oss-20b

Hugging Face (openai) Open weights

The catch: Text only and must use OpenAI's harmony prompt format; the card says it runs within 16GB of memory.

AccessOpen weights
Free limitsFree to download and self-host; you pay only for your own compute. Apache 2.0, so no user cap. 21B total parameters, 3.6B active (mixture of experts).
Modalitytext, code
Credit cardNot required
Commercial useAllowed
Trains on your data–
Licence or termsApache-2.0
Context window131,072 tokens
Model idopenai/gpt-oss-20b
Base URL–
Last checked2 Oct 2026

How to use gpt-oss-20b

Download the weights and serve them yourself; the quickstart below shows one way. Licence: Apache-2.0.

Quickstart
ollama run gpt-oss:20b
Paste into Claude Code
I want to run the open-weight model "gpt-oss-20b" (openai/gpt-oss-20b) for my project.
Please:
1. Check the model card and tell me the hardware it needs and the simplest way to serve it (vLLM, SGLang, Ollama or a hosted endpoint).
2. Give me a minimal working setup and a test request.
3. Tell me the licence terms for commercial use.
4. If a cheaper or smaller variant would do the job, mention it.

Before you ship on it

  • Text only and must use OpenAI's harmony prompt format; the card says it runs within 16GB of memory.
  • Apache 2.0 (HF repo LICENSE). The repo's USAGE_POLICY only asks users to comply with all applicable law.
  • Free to run means you pay for the GPUs, the serving stack and the on-call.

Questions

Is gpt-oss-20b free?

Yes. Open weights offers it as open weights, with these limits: Self-host free · Apache-2.0. No credit card is needed.

Can I use gpt-oss-20b commercially?

Yes, under Apache-2.0. Check the current terms before you ship.

How do I start with gpt-oss-20b?

Download the weights and serve them yourself; the quickstart below shows one way. Licence: Apache-2.0.

Source: huggingface.co/openai/gpt-oss-20b, checked 2 Oct 2026.