This model was built with gpt-oss-120b model by applying AMD-Quark for MXFP4 quantization.
The model was quantized from openai/gpt-oss-120b using AMD-Quark. The weights are quantized MXFP4 and activations were quantized to FP8.
Quantization scripts:
cd Quark/examples/torch/language_modeling/llm_ptq/
exclude_layers="*lm_head *self_attn* *router*"
python3 internal_scripts/quantize_quark.py \
--model_dir openai/gpt-oss-120b \
--quant_scheme w_mxfp4_a_fp8 \
--exclude_layers $exclude_layers \
--num_calib_data 512 \
--output_dir amd/gpt-oss120b-w-mxfp4-a-fp8 \
--model_export hf_format \
--multi_gpu
This model can be deployed efficiently using the vLLM backend.
The model was evaluated on AIME25 and GPQA Diamond benchmarks with low reasoning effort.
| Benchmark | gpt-oss-120b | gpt-oss120b-w-mxfp4-a-fp8(this model) | Recovery |
| AIME25 | 65.25 | 67.12 | 102.87% |
| GPQA | 51.67 | 53.42 | 103.39% |
The results of AIME25 and GPQA Diamond were obtained using gpt_oss.evals with low effort setting, and vLLM docker rocm/vllm-private:rocm7.0_ubuntu_22.04_vllm_0.10.1_instinct_gptoss_wmxfp4_afp8_20251030.
vllm serve amd/gpt-oss120b-w-mxfp4-a-fp8 \
--tensor_parallel_size 2 \
--gpu-memory-utilization 0.90 \
--no-enable-prefix-caching \
--max-num-batched-tokens 1024
python -m gpt_oss.evals --model /shareddata/amd/gpt-oss120b-w-mxfp4-a-fp8 --eval aime25,gpqa --reasoning-effort low --n-threads 128
Modifications Copyright(c) 2025 Advanced Micro Devices, Inc. All rights reserved.
model-00001-of-00014.safetensors 5.0 GBmodel-00011-of-00014.safetensors 5.0 GBmodel-00005-of-00014.safetensors 5.0 GBmodel-00012-of-00014.safetensors 5.0 GBmodel-00006-of-00014.safetensors 5.0 GBmodel-00007-of-00014.safetensors 5.0 GBmodel-00008-of-00014.safetensors 5.0 GBmodel-00009-of-00014.safetensors 5.0 GBmodel-00004-of-00014.safetensors 5.0 GBmodel-00002-of-00014.safetensors 5.0 GBmodel-00010-of-00014.safetensors 5.0 GBmodel-00003-of-00014.safetensors 5.0 GBmodel-00013-of-00014.safetensors 4.2 GBmodel-00014-of-00014.safetensors 1.2 GBHave the files? Submit a torrent.
Submit a torrentHave the complete files and permission to share them? Replace the example folder path. The script checks the files, makes a torrent and submits its magnet and checksums. Keep your torrent client seeding. Get your community key from your settings, and read the script before running it.
curl -fsSL https://pirateface.co/package.sh | bash -s -- --repo jamiefutch/gpt-oss-120b-w-mxfp4-a-fp8 ./model-folderopenai/gpt-oss-120b