OpenAI · First MLX build 29 Aug 2025 · In Core AI since 8 Jun 2026
gpt-oss-20b
MLXCore AI
Text apache-2.0 Trending
A 21B parameter text generation model from OpenAI, with 4 MLX builds, from MXFP4 to 4-bit and Apple Core AI export recipes for macOS.
- Parameters
- 21B From Hugging Face
- Quantisation
- MXFP4 gpt-oss-20b-MXFP4-Q8
- Memory needed
- 16 GB Estimate, unified memory
- Download
- 11.8 GB Weights, estimated
- Downloads
- 173K This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 4 On mlx-community
- Devices
- Mac Apple silicon
- Parameters
- 21B From Hugging Face
- Compression
- None Apple’s preset
- Memory needed
- — Parameter count not published
- Context
- 32K tokens Fixed at export
- Precision
- BF16 Compute precision
- Platforms
- macOS Mac
- Export presets
- 1 In Apple’s registry
- Listed by Apple
- 8 Jun 2026 First registry commit
Memory
Will it fit?Needs about 16 GB at MXFP4
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · mlx-lm$ pip install -U mlx-lm
$ mlx_lm.generate --model mlx-community/gpt-oss-20b-MXFP4-Q8 --prompt "Hello"Memory
There is no memory estimate: the parameter count for this model is not published on Hugging Face.
Export it
Terminal · uv$ git clone https://github.com/apple/coreai-models
$ cd coreai-models
$ uv run coreai.llm.export gpt-oss-20b
Apple publishes the recipe, not a ready-made package: these commands convert the original weights to a Core AI .aimodel on your Mac. See the repository for requirements.
MLX builds
4 on mlx-community| gpt-oss-20b-MXFP4-Q8 | MXFP4 | 11.8 GB | 16 GB | 173K |
|---|---|---|---|---|
| gpt-oss-20b-MXFP4-Q4 | MXFP4 | 11.8 GB | 16 GB | 2.4K |
| gpt-oss-20b-OptiQ-4bit | 4-bit | 11.8 GB | 16 GB | 1.8K |
| gpt-oss-20b-mxfp4-bf16 | MXFP4 | 14.4 GB | 19 GB | 454 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.
Core AI export presets
From Apple’s registry| Preset | Platform | Compression | Precision | Context |
|---|---|---|---|---|
| gpt-oss-20b | macOS | None | BF16 | 32K |