Rent a GPU

DeepSeek V4 Pro 1.6T-A49B hardware requirements

1600B parameters (49B active, mixture of experts). Needs about 971.2 GB at Q4_K_M with 8K context.

Parameters1600B49B active (MoE)
Q4_K_M · 8K971.2 GBthe usual download
FP16 · 8K3201.2 GBfull precision
Runs well on0/51GPUs and Macs we track

Memory needed at each quantization

Weights, KV cache and runtime overhead, at 8K and 32K tokens of context.

QuantQuality8K context32K context
FP16Full quality3201.2 GB3202.9 GB
Q8_0Near-lossless1701.2 GB1702.9 GB
Q6_KExcellent1313.2 GB1314.9 GB
Q5_K_MVery good1139.2 GB1140.9 GB
Q4_K_MGood, most popular971.2 GB972.9 GB
Q3_K_MNoticeably degraded783.2 GB784.9 GB
Q2_KHeavily degraded601.2 GB602.9 GB

Model card

deepseek-ai/DeepSeek-V4-Pro-0813 on Hugging Face: licence, prompt format and GGUF builds.

Run it in the cloud

Run DeepSeek V4 Pro 1.6T-A49B on a rented GPU today.

It needs about 971.2 GB at Q4_K_M: a rented 8× H200 runs it for about $42.32 an hour. Compare with Claude and APIs or see how it scores against them.

RunPodRent a GPU on RunPodVast.aiRent a GPU on Vast.ai

Every GPU and Mac checked

Sorted by memory. Bars show how much of each card DeepSeek V4 Pro 1.6T-A49B fills at the best quant that fits.

51 Too big

NameMemory usedBest quantSpeed

Other DeepSeek models