DeepSeek V4.1 Flash has 763.2B parameters (16B active per token). In BF16 its weights alone take 1395 GiB; quantized to 4-bit, about 421 GiB. Each 32K-token request adds 0.09 GiB of KV cache. Cheapest way to run it today: 7 × NVIDIA A100 80GB at $6.54 per hour.
Read every day from Hugging Face. Split files are added together; vision projectors are left out.
How many GPUs to run DeepSeek V4.1 Flash
Fewest GPUs that fit the weights plus one 8K-token request. Data-center GPUs use vLLM (90% of memory usable), the others llama.cpp. Several GPUs split the model with tensor or pipeline parallelism.
GPU
FP4 experts + FP8 (native)
BF16 / FP16
FP8 / INT8
4-bit (GGUF Q4_K_M)
GeForce RTX 4090 24 GB
24 (several servers)
96 (several servers)
48 (several servers)
24 (several servers)
GeForce RTX 5090 32 GB
24 (several servers)
64 (several servers)
32 (several servers)
16 (several servers)
RTX PRO 6000 Blackwell 96 GB
6
16 (several servers)
8
5
Mac M5 Max 128 GB 96 GB usable
6
16 (several servers)
8
5
NVIDIA H100 80 GB
7
24 (several servers)
16 (several servers)
7
NVIDIA H200 141 GB
4
16 (several servers)
6
4
NVIDIA B200 (HGX) 180 GB
3
16 (several servers)
5
3
Cheapest way to run DeepSeek V4.1 Flash today
For each GPU: the fewest cards that fit DeepSeek V4.1 Flash with one 8K-token request, the most faithful weight format at that count, and the cheapest on-demand price (2026-10-10).
On-demand prices from RunPod, Vast.ai, Verda and Azure, checked every hour. Serving many users needs more KV cache, so more memory: size it in the calculator. Some links are affiliate links: StudioTV may earn a commission, at no extra cost to you.
KV cache: how memory grows with context
Architecture: MoE 384 experts · global KV written by 4 of 40 layers and shared by the rest (890 bytes per token in native FP4) + 128-token window · 196B Engram memory. Its KV cache is 13.4× smaller than if every layer used full attention, because it keeps only a short window of tokens at full detail and compresses the rest of the sequence. Only 16B parameters are active per token, but all 763B must sit in memory: VRAM depends on the total, speed on the active part.
Context per request
BF16 cache
FP8 cache
8K tokens
0.03 GiB
0.02 GiB
32K tokens
0.09 GiB
0.06 GiB
128K tokens
0.36 GiB
0.22 GiB
1M tokens
2.83 GiB
1.75 GiB
Plus 0.02 MiB of recurrent state per request, whatever the context.
DeepSeek V4.1 Flash VRAM FAQ
How much VRAM does DeepSeek V4.1 Flash need?
In BF16 the weights alone take 1395 GiB (1498 GB). In FP8 that is 699 GiB, and about 421 GiB with 4-bit quantization (Q4_K_M); the official FP4 experts + FP8 checkpoint is 468 GiB. Each request then adds KV cache: 0.03 GiB at 8K tokens and 0.09 GiB at 32K (BF16 cache).
Can DeepSeek V4.1 Flash run on a single RTX 4090 (24 GB)?
No. Even in 4-bit it needs 24 (several servers) RTX 4090s. In 4-bit it needs 7 H100 GPUs.
What is the cheapest way to run DeepSeek V4.1 Flash?
On 2026-10-10, the cheapest on-demand setup is 7 × NVIDIA A100 80GB with FP4 experts + FP8 weights on Vast.ai, at $6.54 per hour (about $4,772.74 per month). Next: 5 × RTX PRO 6000 Blackwell (96 GB) with 4-bit (GGUF Q4_K_M) weights on Vast.ai, at $6.67 per hour. Sized for one 8K-token request; prices are checked every hour.
How many H100 GPUs does DeepSeek V4.1 Flash need?
With one request and an 8K-token context: 24 (several servers) in BF16, 16 (several servers) in FP8 and 7 in 4-bit (vLLM, 90% of memory usable). Serving many users at once needs more memory for their KV caches: the calculator sizes that for you.
DeepSeek V4.1 Flash VRAM badge
For a model card or a README: the 4-bit size and the smallest GPU it fits on, linked to this page. Data as JSON: /api/models/deepseek-v4-1-flash.json.
Estimates, not guarantees: computed from the official config.json with the same engine as the LLM VRAM Calculator. Real usage depends on your engine version and settings.
DeepSeek, China · Released September 10, 2026