What GPU do you need to run Llama 4 Maverick?
400B params 128K max context VisionTool use
The smallest GPU memory that runs Llama 4 Maverick well is 256 GB (Q4_K_M, e.g. M3 Ultra). 2 of 132 GPUs run it well.
MMLU 85.5
VRAM needed per quantization
| Quantization | Weights | 16K context | 32K context | Full 128K |
|---|---|---|---|---|
| Q4_K_M | 243.0 GB | 245.9 GB | 248.9 GB | 266.4 GB |
| Q8_0 | 426.0 GB | 428.9 GB | 431.9 GB | 449.4 GB |
Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.
GPUs that can run Llama 4 Maverick
Best quantization that fits on each GPU, with the context and speed it gets. For chips with several memory sizes, the smallest that works is shown.