OpenAI
GPT-OSS 120B (117B parameters) requires approximately 77.8 GB of VRAM with Q4_K_M quantization. For the best balance of quality and speed, we recommend hardware with at least 90 GB of VRAM.
Get started
— copy & paste to run locallyCopy-paste commands to run GPT-OSS 120B on your machine.
Run
ollama run gpt-oss:120bQuick specs
About this model
Related models
Quick picks
Best hardware
Run this model
Quantization options
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 45.6 GB | Low | — |
Q3_K_S | 3 | 57.3 GB | Low | — |
NVFP4 | 4 | 65.5 GB | Medium | — |
Q4_K_M | 4 | 71.4 GB | Medium | — |
Q5_K_M | 5 | 84.2 GB | High | — |
Q6_K | 6 | 95.9 GB | High | — |
Q8_0 | 8 | 125.2 GB | Very High | — |
F16 | 16 | 239.8 GB | Maximum | — |
Quality benchmarks
Coding
Reasoning
Source: official · 2025-08-15
Hardware compatibility
Computing compatibility...
Memory breakdown
Frequently asked questions
GPT-OSS 120B (117B parameters) requires approximately 77.8 GB of VRAM with Q4_K_M quantization. Lower quantizations like Q4_K_M use less memory but may reduce quality.
Yes, Mac Studio M3 Ultra 256GB can run GPT-OSS 120B with a compatibility score of 87/100. It provides 256 GB of memory and achieves approximately 8.5 tokens per second.
The recommended quantization for GPT-OSS 120B is Q4_K_M, which offers the best balance between model quality and memory efficiency. Higher quantizations preserve more quality but require more VRAM.
The top recommended hardware for GPT-OSS 120B: AMD Instinct MI300A 128GB (score: 96/100), NVIDIA H200 141GB (score: 95/100), NVIDIA H200 PCIe 141GB (score: 95/100). These provide the best combination of memory, bandwidth, and compute for running this model locally.
Yes, GPT-OSS 120B is well-suited for chat as well as reasoning, coding. It was designed with these use cases in mind.
See also
