What Models?

Pick the right model for your GPU — in seconds.

What GPU do you need to run Llama 3.2 Vision 11B?

10.6B params 128K max context VisionTool use

The smallest GPU memory that runs Llama 3.2 Vision 11B well is 8 GB (Q4_K_M, e.g. GeForce RTX 5060 Ti). 120 of 132 GPUs run it well.

MMLU 73

VRAM needed per quantization

QuantizationWeights16K context32K contextFull 128K
Q4_K_M6.0 GB7.9 GB9.9 GB21.6 GB
Q8_010.4 GB12.4 GB14.3 GB26.0 GB

Weights plus fp16 KV cache. Leave ~1 GB headroom for the inference engine.

Cheapest GPUs that can run Llama 3.2 Vision 11B

Lowest launch MSRP among GPUs that run its smallest quantization (Q4_K_M) well, with at least 16K context.

As an Amazon Associate, I earn from qualifying purchases.

GPUs that can run Llama 3.2 Vision 11B

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.

NVIDIA

AMD

Intel

Apple

  • M1 8 GB Q4_K_M 16K context · ~10 tok/s
  • M2 8 GB Q4_K_M 16K context · ~14 tok/s
  • M3 8 GB Q4_K_M 16K context · ~14 tok/s
  • M1 Pro 16 GB Q8_0 45K context · ~18 tok/s
  • M2 Pro 16 GB Q8_0 45K context · ~18 tok/s
  • M4 16 GB Q8_0 45K context · ~11 tok/s
  • M5 16 GB Q8_0 45K context · ~13 tok/s
  • M6 16 GB Q8_0 45K context · ~13 tok/s
  • M3 Pro 18 GB Q8_0 62K context · ~13 tok/s
  • M4 Pro 24 GB Q8_0 111K context · ~24 tok/s
  • M5 Pro 24 GB Q8_0 111K context · ~27 tok/s
  • M1 Max 32 GB Q8_0 128K context · ~35 tok/s
  • M2 Max 32 GB Q8_0 128K context · ~35 tok/s
  • M3 Max 36 GB Q8_0 128K context · ~26 tok/s
  • M4 Max 36 GB Q8_0 128K context · ~36 tok/s
  • M5 Max 36 GB Q8_0 128K context · ~40 tok/s
  • M1 Ultra 64 GB Q8_0 128K context · ~70 tok/s
  • M2 Ultra 64 GB Q8_0 128K context · ~70 tok/s
  • M3 Ultra 96 GB Q8_0 128K context · ~72 tok/s
  • M5 Ultra 96 GB Q8_0 128K context · ~105 tok/s
Check your exact setup in the full tool →