pd95/ gptoss-mlx:20b-mxfp4-r0.30.10-20260622

365 1 week ago

Experimental MLX-backed GPT-OSS 20B for custom Ollama builds. Requires GptOssForCausalLM support and is not compatible with the regular Ollama app.

tools thinking 20b 
ollama run pd95/gptoss-mlx:20b-mxfp4-r0.30.10-20260622

Details

1 week ago

50d14eb2f9db · 12GB ·

json
{ "architectures": [ "GptOssForCausalLM" ], "attention_bias": true, "attention_dropout": 0.0, "eos_t
json
{ "bos_token_id": 199998, "do_sample": true, "eos_token_id": [ 200002, 199999, 200012 ], "pad_token_
json
{ "bos_token": "<|startoftext|>", "eos_token": "<|return|>", "pad_token": "<|endoftext|>" }
json
{ "version": "1.0", "truncation": null, "padding": null, "added_tokens": [ { "id": 199998, "content"
json
{ "added_tokens_decoder": { "199998": { "content": "<|startoftext|>", "lstrip": false, "normalized":
license
Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR US
params
{ "temperature": 1 }
template
<|start|>system<|message|>You are ChatGPT, a large language model trained by OpenAI. Knowledge cutof
339 tensors

Readme

gptoss-mlx

GPT-OSS 20B for the experimental Ollama MLX runner.

This model requires a custom Ollama build with experimental MLX safetensors and GptOssForCausalLM support. It does not run with the regular public Ollama app.

The model declares Ollama 0.31.1 as its minimum version. This refers to the custom MLX preview series—not the official Ollama 0.31.1 build. The currently recommended build is available at:

https://www.doapp.ch/Ollama/

Tags

  • pd95/gptoss-mlx:20b — recommended moving alias
  • pd95/gptoss-mlx:20b-mxfp4 — explicit MXFP4 moving alias
  • pd95/gptoss-mlx:20b-mxfp4-v0.32.15-r2 — immutable release artifact

All three tags currently reference the same model manifest.

Run the recommended tag with:

”`console ollama run pd95/gptoss-mlx:20b

Compatibility

This artifact was created with Ollama MLX preview v0.32.15-r2.

It has been tested successfully with custom MLX preview runtimes from v0.31.1 through v0.32.15-r2. Older previews are not supported.

Source and provenance

Upstream model:

https://huggingface.co/openai/gpt-oss-20b

Immutable source revision:

6cee5e81ee83917806bbde320786a8fb61efebee

Publication details:

  • Source architecture: GPT-OSS 20B
  • Source quantization: native MXFP4
  • Conversion: Ollama safetensors import with the upstream MXFP4 tensor representation retained
  • Import preview: v0.32.15-r2
  • Minimum declared Ollama version: 0.31.1
  • Published manifest SHA-256: 99d41238eeeb1f6e66e4ad0e4e4cc4299b796977aa822353e68dbeddf699e1ec

The moving 20b and 20b-mxfp4 tags supersede the earlier published artifact. Use the immutable 20b-mxfp4-v0.32.15-r2 tag when reproducibility is important.

Capabilities

  • Text generation
  • Tool calling
  • Thinking
  • 128K context window
  • Experimental Ollama MLX runtime

License and use policy

GPT-OSS is released by OpenAI under the Apache License 2.0:

https://huggingface.co/openai/gpt-oss-20b/blob/main/LICENSE

Use is also subject to the upstream model terms and safety guidance:

https://huggingface.co/openai/gpt-oss-20b

This repository republishes an MLX/Ollama artifact. It does not modify the upstream license or usage terms. Review the upstream license and model card before use or redistribution.