Models · family
GPT-OSS
1 MLX alias · OpenAI's GPT-OSS 20B in mxfp4-q8 · Harmony-native tool calling.
OpenAI's GPT-OSS 20B in MLX form — Harmony-native tool calling (use the harmony tool-call parser) and Harmony reasoning streams. mxfp4-q8 is the recommended-by-OpenAI low-bit format and runs comfortably on a 32 GB Mac.
- family
- GPT-OSS
- aliases
- 1
- lines
- 1
- install
- rapid-mlx serve <alias>
- OpenAI base URL
- http://localhost:8000/v1
GPT-OSS 20B · 1 alias
mxfp4-q8 — the recommended-by-OpenAI low-bit format. Harmony-native tool calling + Harmony reasoning streams.
parser: harmony
| alias | hf repo | tool parser | reasoning | flags |
|---|---|---|---|---|
| gpt-oss-20b-mxfp4-q8 | mlx-community/gpt-oss-20b-MXFP4-Q8 | harmony | harmony | moe · spec |
Notes & caveats
- 100% tool-call format compliance — Harmony was designed end-to-end for tools.
- Reasoning streams via the
harmonyreasoning parser intoreasoning_contenton/v1/responses.