>>

Run DeepSeek-V4-Flash-Vision-Exp on Lium

DeepSeek · Vision and text · 304.6B parameters · 167.8 GB 4-bit weights · released 2026-08-31
It needs about 202 GB of VRAM. The cheapest setup that runs it is 8x RTX 5090 (256 GB), from $4.00 per hour, billed per second.

Other GPUs that run it

8x RTX 6000 Ada Generation384 GBfrom $6.00 / hourRent
2x H200282 GBfrom $8.40 / hourRent

How to run it

Rent the pod above, open a terminal over SSH or Jupyter and serve deepseek-ai/DeepSeek-V4-Flash-Vision-Exp:
vllm serve deepseek-ai/DeepSeek-V4-Flash-Vision-Exp --tensor-parallel-size 8
Claude Code and Codex can do all of it through the Lium CLI.

Frequently asked questions

How much VRAM does DeepSeek-V4-Flash-Vision-Exp need?

About 167.8 GB for the released 4-bit weights, so plan on 202 GB with room for activations and KV cache at a short context. Longer contexts and bigger batches need more.

What GPU should I rent to run DeepSeek-V4-Flash-Vision-Exp?

8x RTX 5090 (256 GB). On Lium that setup costs from $4.00 per hour, billed per second with no minimum.
All open models
For AI agents (CLI install, agent skill, headless signup): llms-full.txt