gpt-oss 20B
browserOpenAI's open model. Cleanest local driver for the web browser tool. Needs MXFP4 (the only quant gpt-oss runs well in).
Parameters20.0B
Context131,072 tokens
Download12.1 GB
Min RAM16 GB
Architecturegpt-oss
Available quantizations
| Quant | Size | Quality |
|---|---|---|
| MXFP4 | 12.1 GB | best |
Check whether your machine can run this model — with real measured speeds from anonymous community benchmarks — on the Central-Intel compatibility page.