How much memory does gpt-oss 120B need?
About 73 GB at Q4_K_M with an 8K context. It runs on a Mac with 128 GB of unified memory or more.
Try other settings in the calculatorMemory by quantization and context
Total memory: weights, KV cache in F16 and about 1 GB of buffers. Under each total, the smallest Mac that can run it with the default GPU memory limit.
| Quantization | 8K context | 32K context | 128K context |
|---|---|---|---|
| Q3_K_M | 58 GB96 GB Mac | 59 GB96 GB Mac | 63 GB96 GB Mac |
| Q4_K_M | 73 GB128 GB Mac | 74 GB128 GB Mac | 77 GB128 GB Mac |
| Q5_K_M | 85 GB128 GB Mac | 85 GB128 GB Mac | 89 GB128 GB Mac |
| Q6_K | 97 GB192 GB Mac | 98 GB192 GB Mac | 102 GB192 GB Mac |
| Q8_0 | 125 GB192 GB Mac | 126 GB192 GB Mac | 130 GB192 GB Mac |
| F16 | 235 GBtoo big for a Mac | 236 GBtoo big for a Mac | 239 GBtoo big for a Mac |
About the model
- Parameters
- 117B
- Type
- Mixture of experts, 128 experts
- Attention
- sliding window plus full attention
- KV cache, F16
- 0.3 GB at 8K, 4.8 GB at 128K
- On Hugging Face
- openai/gpt-oss-120b
Some layers only look at the last 128 tokens, so their part of the cache stops growing. The full-attention layers still grow with context. As a mixture-of-experts model it loads every expert into memory, so memory follows the total parameter count, while speed follows the smaller number of active parameters.
Questions
How much memory does gpt-oss 120B need?
About 73 GB at Q4_K_M with an 8K context: 71 GB of weights, 0.3 GB of KV cache and about 1 GB of buffers. At 32K context it needs about 74 GB.
Can gpt-oss 120B run on a Mac?
Yes. At Q4_K_M with an 8K context it fits on a Mac with 128 GB of unified memory or more, using the default GPU memory limit of about three quarters.
How big is the gpt-oss 120B KV cache?
About 0.3 GB at 8K tokens and 4.8 GB at 128K tokens with an F16 cache. Its attention design: sliding window plus full attention.
Other models
- Gemma 4 12B
- gpt-oss 20B
- Gemma 4 26B-A4B
- Qwen3.8 27B
- GLM-4.7 Flash
- Gemma 4 31B
- Qwen3.6 35B-A3B
- Mistral Medium 3.5
- Qwen3.8 Flash Next
- DeepSeek V4 Flash
- GLM-5.3
- DeepSeek V4 Pro
- Kimi K3
Data read from Hugging Face on 2026-11-09. These are estimates; benchmarks on this site report measured memory. How the estimate works. Compare all models in the model fit chart.
Get the next one
New benchmarks and agent runs by email, with the raw data. Or follow on RSS or X.