# DeepSeek-V4-Flash GPU requirements

DeepSeek-V4-Flash has 291B parameters. To serve it, plan for about 698 GB of GPU memory in 16-bit, 349 GB in 8-bit or 175 GB in 4-bit, including about 20% for the KV cache and activations.

It is a mixture of experts: 256 experts, 6 of them active for each token. All of them stay in GPU memory, so plan memory for all 291B parameters.

[Deploy it on Arrakis](/console/#/models?model=deepseek-ai/DeepSeek-V4-Flash).

## What it takes to serve it

| Precision | Memory and the smallest setup |
|---|---|
| 16-bit | 698 GB: 4 × [B200](/gpus/b200.html.md), from $27.16 per hour for all 4 |
| 8-bit | 349 GB: 2 × [B200](/gpus/b200.html.md), from $13.58 per hour for all 2 |
| 4-bit | 175 GB: one [B200](/gpus/b200.html.md), from $6.79 per GPU-hour |

## What it takes to fine-tune it

| Method | Memory and the smallest setup |
|---|---|
| QLoRA | 204 GB: one [B300](/gpus/b300.html.md), from $4.60 per GPU-hour |
| LoRA | 698 GB: 4 × [B200](/gpus/b200.html.md), from $27.16 per hour for all 4 |
| Full fine-tuning | 4,655 GB: more than one 8-GPU machine |

## About the model

| | |
|---|---|
| Parameters | `291B` |
| Experts | `256` |
| Experts for each token | `6` |
| Context window (tokens) | `1,048,576` |
| Published in | `FP8` |
| License | `mit` |
| Released | `2026-04-22` |
| Weights | [deepseek-ai/DeepSeek-V4-Flash](https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash) |

## Serve it yourself

```
vllm serve deepseek-ai/DeepSeek-V4-Flash --tensor-parallel-size 4
```

## Questions

**How much GPU memory does DeepSeek-V4-Flash need?** To serve it, about 698 GB in 16-bit, 349 GB in 8-bit, 175 GB in 4-bit, including about 20% for the KV cache and activations. To fine-tune it, about 204 GB with QLoRA, 698 GB with LoRA and 4,655 GB for full fine-tuning with Adam, before activations.

**Can I run DeepSeek-V4-Flash on one GPU?** Yes, quantized. In 4-bit it needs about 175 GB, which fits on one B200. In 16-bit it needs 4 B200 GPUs.

**What does it cost to run DeepSeek-V4-Flash?** In 16-bit it needs 4 B200 GPUs; in 4-bit, one B200. Each setup on this page shows the lowest price per GPU-hour on Arrakis right now, and you pay by the hour from a prepaid balance.

**How do I deploy DeepSeek-V4-Flash on Arrakis?** Open Models in the Arrakis console, paste deepseek-ai/DeepSeek-V4-Flash, choose vLLM, SGLang or llama.cpp, and rent a GPU it fits on. You get an OpenAI-compatible endpoint with its own API key, and root SSH to the machine.


More models: [/models](/models.html.md).