The Jetson T5000 only has 128GB VRAM. Llama 3.1 405B needs ~246.5GB (estimate). Time to rent or upgrade.
Estimate · Q4_K_M · 4k context · batch 1. For dense 7B–70B this sits ~5–8% above the Q4_K_M GGUF file. Not peak runtime VRAM and not a measured bench.
Need more VRAM?
Rent cloud GPUs with 246.5GB+ VRAM to run Llama 3.1 405B, or upgrade hardware.
This GPU is short by ~118.5GB VRAM for Llama 3.1 405B at Q4.
We may earn a commission from cloud and hardware partners at no extra cost to you.