DeepSeek V4 Flash 0731 GPU requirements: VRAM and cheapest GPU
DeepSeek V4 Flash 0731 has 304.2B parameters, published in FP4. It is a mixture-of-experts model: 6 of its 256 experts work on each token, about 33.6B parameters, so it generates at roughly the speed of a model that size. Every expert still has to sit in GPU memory, which is why the memory figures below are for the full model. It is released under mit, which allows commercial use.
- Parameters
- 304.2B
- Smallest footprint (INT4)
- about 177 GB
- Cheapest GPU per hour (FP16)
- $2.39
- Type
- Text
deepseek-ai/DeepSeek-V4-Flash-0731
- Parameters 304.2B
- Active per token 33.6B
- Type Text
- Task text-generation
- Max context 1024K
- Licence mit
- Downloads 2.4M
- Likes 4.1K
Calculating for DeepSeek V4 Flash 0731
More settings
DeepSeek V4 Flash 0731 needs about 671 GB of GPU memory
Inference (vLLM), FP16 / BF16, 8K tokens context, one request at a time.
Where the memory goes
- Weights 608 GB
- KV cache 0.72 GB
- Runtime overhead 62 GB
- Total 671 GB
GPU counts keep 10% of each card free for spikes. Plan extra if you expect long prompts or many users at once.
| Rank | Configuration | Per GPU / hr | GPUs needed | Total / hr | Total / month (24/7) | Provider |
|---|---|---|---|---|---|---|
| 1 | 4 × MI300X 192GBCheapestCDNA 3 · HBM3 | $2.39 | 4 | $9.56 | $6,979 | RunPod (opens in a new tab)4x MI300X |
| 2 | 8 × RTX PRO 6000 96GBBlackwell · GDDR7 | $1.80 | 8 | $14.40 | $10,512 | Nebius (opens in a new tab)8x RTX PRO 6000 96GB |
| 3 | 8 × H100 NVL 94GBHopper · HBM3 | $3.19 | 8 | $25.52 | $18,630 | RunPod (opens in a new tab)8x H100 NVL |
| 4 | 6 × H200 141GBHopper · HBM3e | $3.79 | 6sold as 8 GPUs; you need 6 | $30.32 | $22,134 | RunPod (opens in a new tab)8x H200 NVL |
| 5 | 3 × B300 288GBBlackwell Ultra · HBM3e | $8.99 | 3sold as 4 GPUs; you need 3 | $35.96 | $26,251 | RunPod (opens in a new tab)4x B300 |
| 6 | 5 × B200 180GBBlackwell · HBM3e | $7.69 | 5sold as 8 GPUs; you need 5 | $61.52 | $44,910 | Lambda (opens in a new tab)8x HGX B200 |
| More than 8: needs several nodes: H100 80GB, A100 80GB, L40S 48GB, RTX 6000 Ada 48GB, RTX A6000 48GB, A40 48GB, A100 40GB, RTX 5090 32GB, RTX 4090 24GB, RTX 3090 24GB, L4 24GB, A10 24GB, T4 16GB, V100 16GB | ||||||
| Provider | Instance | Per GPU / hr | GPUs needed | Total / hr | Total / month (24/7) |
|---|---|---|---|---|---|
| RunPod (opens in a new tab)Cheapest | 4x MI300Xus | $2.39 | 4 × MI300X 192GB | about $9.56 | about $6,979 |
| Nebius (opens in a new tab) | 8x RTX PRO 6000 96GBeu/us | $1.80 | 8 × RTX PRO 6000 96GB | about $14.40 | about $10,512 |
| CoreWeave (opens in a new tab) | RTX PRO 6000 Blackwell Server Edition (High Memory)us | $2.50 | 8 × RTX PRO 6000 96GB | about $20.00 | about $14,600 |
| Vultr (opens in a new tab) | vbm-256c-2048gb-8-mi300x-gpu (bare metal)us | $3.99 | 4 × MI300X 192GB sold as 8 GPUs; you need 4 | about $31.92 | about $23,302 |
| AWS (opens in a new tab) | g7e.48xlargeus-east-1 | $4.14 | 8 × RTX PRO 6000 96GB | about $33.14 | about $24,195 |
| Google Cloud (opens in a new tab) | g4-standard-384us-central1 | $4.50 | 8 × RTX PRO 6000 96GB | about $36.00 | about $26,280 |
| Lambda (opens in a new tab) | 8x HGX B200us | $7.69 | 5 × B200 180GB sold as 8 GPUs; you need 5 | about $61.52 | about $44,910 |
Prices collected Nov 13, 2026. Every cost is approximate: on-demand, US dollars, mostly US regions (marketplaces such as Vast.ai are worldwide).
Memory by precision and workload
| Precision | Inference | LoRA fine-tune | QLoRA fine-tune | Full fine-tune |
|---|---|---|---|---|
| FP16 / BF16 | 671 GB | 782 GB | – | 5,359 GB |
| INT8 | 353 GB | – | – | – |
| INT4 | 177 GB | – | 288 GB | – |
Cheapest GPU by precision
Inference with an 8K context and one request at a time.
FP16 / BF16
about 671 GB · 8K tokens
The weights as published. Nothing lost, but the largest footprint.
4 × MI300X 192GB
RunPod (opens in a new tab) · 4x MI300X
- Per GPU / hr
- $2.39
- GPUs needed
- 4
- Total / hr
- about $9.56
- per month, 24/7, approx.
- $6,979
INT8
about 353 GB · 8K tokens
Weights at half the size. Most chat and RAG work shows no visible difference.
3 × MI300X 192GB
RunPod (opens in a new tab) · 4x MI300X
sold as 4 GPUs; you need 3- Per GPU / hr
- $2.39
- GPUs needed
- 3
- Total / hr
- about $9.56
- per month, 24/7, approx.
- $6,979
INT4
about 177 GB · 8K tokens
Weights at a quarter of the size. Small drops show up first in maths and code.
5 × A40 48GB
RunPod (opens in a new tab) · 8x A40
sold as 8 GPUs; you need 5- Per GPU / hr
- $0.59
- GPUs needed
- 5
- Total / hr
- about $4.72
- per month, 24/7, approx.
- $3,446
Inference vs fine-tuning
Inference (FP16)
1×
of the memory inference needs
671 GB
4 × MI300X 192GB · about $9.56
LoRA fine-tune
1.2×
of the memory inference needs
782 GB
5 × MI300X 192GB · about $19.12
QLoRA fine-tune
0.4×
of the memory inference needs
288 GB
7 × A40 48GB · about $4.72
Full fine-tune
8×
of the memory inference needs
5,359 GB
more than one 8-GPU node
Serving it at FP16 fits on 4 × MI300X 192GB (RunPod). LoRA needs 5 × MI300X 192GB (RunPod), QLoRA 7 × A40 48GB (RunPod) and a full fine-tune more than one 8-GPU node. Training keeps activations and optimiser state in memory as well as the weights, which is why it so often needs more GPUs than serving the same model.
Will it run on a Mac or a single RTX 4090?
RTX 4090 (24 GB)
Not on its own. Even at INT4 it needs about 177 GB, more than the card’s 24 GB. Two or more cards, or a bigger data-centre GPU, will do it.
Mac with Apple silicon
Yes, with enough memory. At INT4 it needs about 177 GB, so a Mac with 256 GB or more of unified memory (macOS lets the GPU use about three quarters of it). Expect slower generation than on a data-centre GPU.
Similar models
Open models of a similar size.
- Hy3Text298.8B · about 661 GB FP16 · 1M ↓GPU needs
- DeepSeek V4 FlashText290.9B · about 642 GB FP16 · 1.2M ↓GPU needs
- Solar Open 2Text250.3B · about 553 GB FP16 · 7.4K ↓GPU needs
- Trinity-Large-ThinkingText398.6B · about 879 GB FP16 · 3.2K ↓GPU needs
- MiniMax M2.5Text228.7B · about 506 GB FP16 · 1.1M ↓GPU needs
- Laguna M.1Text225.8B · about 500 GB FP16 · 728.4K ↓GPU needs
DeepSeek V4 Flash 0731: common questions
How much VRAM does DeepSeek V4 Flash 0731 need?
About 671 GB at FP16, 353 GB at INT8 and 177 GB at INT4, with an 8K context and one request at a time. Longer contexts and more parallel requests add KV cache on top.
What is the cheapest GPU to run DeepSeek V4 Flash 0731?
4 × MI300X 192GB (about $2.39 per GPU / hour) on RunPod, at about $9.56 an hour for FP16 / BF16 inference with an 8K context. Prices collected Nov 13, 2026; they change often.
Can I fine-tune DeepSeek V4 Flash 0731 on a single GPU?
Not on one GPU. Even QLoRA needs about 288 GB, more than an 80 GB GPU holds. LoRA needs about 782 GB and a full fine-tune about 5,359 GB.
How are these numbers calculated?
Weights are parameters times bytes per parameter (2 at FP16, 1 at INT8, 0.5 at INT4). The KV cache comes from the model’s layers, KV heads, head size and context length. We add about 10% plus 1 GB for the runtime and fill each GPU to 90%. Prices are on-demand list prices in US regions, collected from each provider’s public pricing.
Can DeepSeek V4 Flash 0731 run on a consumer GPU like the RTX 4090?
Not on its own. Even at INT4 it needs about 177 GB, more than the card’s 24 GB. Two or more cards, or a bigger data-centre GPU, will do it.
Does DeepSeek V4 Flash 0731 fit on a single H100 80GB?
No. Even at INT4 it needs about 177 GB, so you need several GPUs.
How many GPUs do I need for DeepSeek V4 Flash 0731?
At FP16 with an 8K context: 10 × 80 GB, 16 × 48 GB or 32 × 24 GB GPUs. At INT4 you need about a quarter of the memory for the weights.
How much VRAM does DeepSeek V4 Flash 0731 need at INT8?
About 353 GB: 319 GB for the weights, 0.72 GB for an 8K KV cache and 33 GB of runtime overhead.
How much does it cost to run DeepSeek V4 Flash 0731 24/7 for a month?
About $6,979 on the cheapest setup we found (4 × MI300X 192GB (about $2.39 per GPU / hour), $9.56 an hour for 730 hours). The cheapest on AWS, Google Cloud or Azure is about $24,195 a month.
FP16 vs INT8 vs INT4: which should I use?
FP16 keeps full quality and needs the most memory (about 671 GB here). INT8 halves the weights with little loss for most chat work. INT4 (AWQ, GPTQ or GGUF Q4) cuts them to a quarter, about 177 GB in total, with a small drop in quality that shows most in maths and code.
Can DeepSeek V4 Flash 0731 run on a Mac?
Yes, with enough memory. At INT4 it needs about 177 GB, so a Mac with 256 GB or more of unified memory (macOS lets the GPU use about three quarters of it). Expect slower generation than on a data-centre GPU.
vLLM or llama.cpp?
vLLM is built for serving many requests at once on NVIDIA and AMD GPUs and reserves most of the GPU memory for KV cache. llama.cpp runs GGUF files on almost anything, including Macs and CPUs, and suits one user or a small team. Memory needs are similar for the same precision.
Prices collected Nov 13, 2026. Every cost is approximate: on-demand, US dollars, mostly US regions (marketplaces such as Vast.ai are worldwide).
How we calculate
- Weights. Parameters times bytes per parameter: 2 at FP16, 1 at INT8, 0.5 at INT4, plus about 5% for the quantisation scales at INT8 and INT4.
- KV cache. 2 × layers × KV heads × head size × context × requests × 2 bytes, from the model’s config.json. Where a setting is missing we estimate it and say so.
- Overhead. About 10% plus 1 GB for the runtime, in line with what vLLM reserves.
- Fine-tuning. LoRA keeps 16-bit weights and adds a small adapter, its optimiser and activations. QLoRA does the same over 4-bit weights. A full fine-tune needs about 16 bytes per parameter with mixed-precision Adam (weights, gradients and optimiser states), plus activations.
- GPUs. We fill each GPU to 90% of its memory and stop at 8 per node. Costs are on-demand list prices in US regions; marketplace clouds show the median offer.