# gpt-oss-120b: hardware requirements
> gpt-oss-120b (OpenAI) is a 117B (5.1B active) mixture-of-experts model with a 128K-token context window, released under the Apache 2.0 license. It needs about 68GB of memory at MXFP4 with 8K context.
gpt-oss-120b is OpenAI's larger open-weight reasoning model: 117B total parameters and 5.1B active, with expert weights shipped natively in MXFP4. It became the default model for 96GB to 128GB unified-memory machines because it is both strong and unusually fast for its size, and OpenAI has not replaced it. The native weights take about 65GB, so it runs fully in memory on an 80GB or 96GB GPU, a Mac with 96GB or more, a Strix Halo 128GB system, or a DGX Spark; a 24GB GPU plus 64GB of system RAM also works with the experts offloaded. Newer models such as Qwen3.5 122B-A10B edge it on benchmarks, but few match its speed.

## Recommended hardware
- **Minimum GPU:** RTX PRO 5000 72GB (72GB), MXFP4, ~145 tokens/sec
- **Best value:** RTX PRO 6000 (96GB), MXFP4, ~170 tokens/sec
- **Fastest single GPU:** RTX PRO 6000 (96GB), MXFP4, ~170 tokens/sec
- **Multi-GPU:** 3× RTX 3090 (72GB), MXFP4, ~66 tokens/sec
- **Multi-GPU:** 3× RTX 4090 (72GB), MXFP4, ~71 tokens/sec
- **Multi-GPU:** 3× RTX 5090 (96GB), MXFP4, ~88 tokens/sec
- **Mac / unified memory:** Ryzen AI Max+ 395 (128GB), MXFP4, ~35 tokens/sec

## Memory by quantization
| Quant | Weights | Total @ 8K ctx | Total @ 32K ctx |
| --- | --- | --- | --- |
| MXFP4 | 65.4GB | 68.3GB | 69.2GB |

## GPUs that run it fully in memory (fastest first)
| GPU | Memory | Quant | Speed | Price |
| --- | --- | --- | --- | --- |
| RTX PRO 6000 | 96GB | MXFP4 | ~170 tok/s | ~$16,000 |
| RTX PRO 5000 72GB | 72GB | MXFP4 | ~145 tok/s | ~$9,100 |
| M5 Ultra 512GB | 512GB | MXFP4 | ~93 tok/s | n/a |
| M5 Ultra 256GB | 256GB | MXFP4 | ~93 tok/s | ~$9,499 |
| M3 Ultra 512GB | 512GB | MXFP4 | ~70 tok/s | $9,499 MSRP |
| M5 Max 128GB | 128GB | MXFP4 | ~64 tok/s | ~$5,100 |
| M4 Max 128GB | 128GB | MXFP4 | ~58 tok/s | $3,499 MSRP |
| DGX Spark | 128GB | MXFP4 | ~37 tok/s | ~$6,950 |
| Ryzen AI Max+ PRO 495 | 192GB | MXFP4 | ~37 tok/s | ~$6,800 |
| Ryzen AI Max+ 395 | 128GB | MXFP4 | ~35 tok/s | ~$3,300 |

## How to run it
- Ollama: `ollama run gpt-oss:120b`
- llama.cpp: `llama-server -hf ggml-org/gpt-oss-120b-GGUF -c 8192 -ngl 99`
- Weights: https://huggingface.co/openai/gpt-oss-120b

## Strengths
- Excellent reasoning, math, and tool calling with only 5.1B active parameters
- Native MXFP4 weights: no quantization decisions and no quality loss
- Small KV cache (36 KiB per token) leaves room for long contexts
- Apache 2.0 license

## Weaknesses
- Text only
- Weaker world knowledge and creative writing than its benchmark scores suggest
- Reasoning is always on, adding latency to simple requests
- Needs about 68GB with context, beyond any single consumer GPU without offload
Full page: https://llamabuilds.ai/models/gpt-oss-120b · Hardware guide: https://llamabuilds.ai/models/gpt-oss-120b/requirements

---
Source: https://llamabuilds.ai/models/gpt-oss-120b · Updated October 14, 2026 · Llama Builds. Speeds marked ~ are estimates from our bandwidth model (https://llamabuilds.ai/methodology); measured results are labeled as such.