gpt-oss 120B
116.8B total, 5.8B active · 128k context · apache-2.0 · released 2025-08-04
chatcodingagenticreasoning
What to run it on in the UK
Fully in memory at 128k context, priced at retailers today.
Something we price holds it fully in memory, but none reaches 15 tok/s, our bar for recommending it. The table below has the speeds.
All hardware that runs it
Everything we price, best value first. Two cards split the model by layers.
| Hardware | Quant | Fits | Decode | Price | tok/s per 1,000 |
|---|---|---|---|---|---|
| Mac Studio M4 Max 128GB | MXFP4 | fits | 27 tok/s | – | – |
| Framework Desktop Ryzen AI Max+ 395 128GB | MXFP4 | fits | 11 tok/s | – | – |
| Mac Studio M3 Ultra 512GB | MXFP4 | fits | 40 tok/s | – | – |
| NVIDIA DGX Spark 128GB | MXFP4 | fits | 13 tok/s | – | – |
| GMKtec EVO-X2 Ryzen AI Max+ 395 128GB | MXFP4 | fits | 11 tok/s | – | – |
| Mac Studio M3 Ultra 256GB | MXFP4 | fits | 40 tok/s | – | – |
| Mac Studio M3 Ultra 96GB | MXFP4 | fits | 40 tok/s | – | – |
| NVIDIA RTX PRO 6000 Blackwell | MXFP4 | fits | 84 tok/s | – | – |
| ASUS Ascent GX10 128GB | MXFP4 | fits | 13 tok/s | – | – |
31 other cards and machines we price cannot hold it at this context.
Memory needed
Weights, KV cache and runtime buffers at each context length (KV cache at fp16).
| Quant | File | @8k | @32k | @64k | @128k |
|---|---|---|---|---|---|
| MXFP4gguf | 59.0 GB | 60.1 GB | 61.1 GB | 62.4 GB | 65.0 GB |
| mlx-4bitmlx | 61.3 GB | 62.3 GB | 63.3 GB | 64.6 GB | 67.2 GB |
Run it
Copy a command for the runtime you use.
- MXFP4
- Ollama
ollama run hf.co/ggml-org/gpt-oss-120b-GGUF:MXFP4 - LM Studio
lms get ggml-org/gpt-oss-120b-GGUF@mxfp4 - llama.cpp
llama-server -hf ggml-org/gpt-oss-120b-GGUF:MXFP4
- Ollama
- mlx-4bit
- LM Studio
lms get mlx-community/gpt-oss-120b-4bit - MLX
mlx_lm.server --model mlx-community/gpt-oss-120b-4bit
- LM Studio