No — not on this device

DeepSeek V4.1 Flash 552B-A16B at Q4_K_M needs 311.0 GB against 44.8 GB usable, and the shortfall of 266.2 GB is more than 32 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.

Does not fit 8K context Q4_K_M · 310.4 GB MIT Released 10 Sep 2026 New this month Vision Not in the Ollama library

Twice V4 Flash's backbone (552B, 16B active per generated token) plus 196B of Engram lookup tables (189 GiB at FP8) that DwarfStar streams from SSD, so they are not counted here. A 512 GB Mac model, and as of October 2026 no mainline llama.cpp or Ollama build: DwarfStar on a Mac, vLLM across four GPUs. DeepSeek puts the cache at 890 bytes/token; modelled conservatively as 4 latent layers plus a 128-token window.

Hardware requirements What hardware do I need to run DeepSeek V4.1 Flash 552B-A16B? →

At Q4_K_M with an 8K context: a 512 GB Mac (M3 Ultra). Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

The VRAM budget

weights 310.4 GB
Weights 310.4 GB KV cache @ 8K 0.04 GB Runtime overhead 0.6 GB Over budget 266.2 GB past 44.8 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 546.2 GB 546.9 GB — ~1.0 −0.1% ppl 502.1 GB over
Q6_K 421.6 GB 422.2 GB — ~1.3 −0.4% ppl 377.4 GB over
Q5_K_M 364.4 GB 365.0 GB — ~1.5 −0.8% ppl 320.2 GB over
Q4_K_M 310.4 GB 311.0 GB — ~1.8 −1.9% ppl 266.2 GB over
Q3_K_M 251.3 GB 251.9 GB — ~2.3 −5.4% ppl 207.1 GB over
Q2_K 215.3 GB 215.9 GB — ~2.8 −15% ppl 171.1 GB over

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model interleaves sliding-window layers (128 tokens, 4 of 40 layers global), which is why its cache barely grows with context. This model uses multi-head latent attention, so its cache is a compressed latent rather than full K and V.

How to run it

terminal
# in the vllm/vllm-openai:deepseekv41-flash-0909 image
$ vllm serve nvidia/DeepSeek-V4.1-Flash-NVFP4 \
    --tensor-parallel-size 4 \
    --tokenizer-mode deepseek_v41 \
    --reasoning-parser deepseek_v41 \
    --language-model-only \
    --max-model-len 8192

The command NVIDIA tested on four GB300 GPUs with its NVFP4 checkpoint, text only. No single card holds it.

A serving engine, not a chat app. Built for many requests at once. More on vLLM.

01The NVFP4 checkpoint is 491 GiB, split across four GPUs by tensor parallelism.
02llama.cpp and Ollama spread the layers over both cards on their own. The cards take turns on every token, so expect about one card's speed; vLLM's --tensor-parallel-size 2 runs them in parallel instead.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on a 2× GeForce RTX 4090

ModelTotalTok/s est.
DeepSeek V4 Flash 284B-A13B
284B MoE · same family
Q4_K_M · 160.7 GB — Won't fit
DeepSeek V4 Pro 1.6T-A49B
1650B MoE · same family
Q4_K_M · 928.9 GB — Won't fit
DeepSeek-R1 671B
671B MoE · similar size
Q4_K_M · 378.4 GB — Won't fit
GLM-5.2 744B-A40B
753B MoE · similar size
Q4_K_M · 424.7 GB — Won't fit
GLM-5.3 744B-A40B
753B MoE · similar size
Q4_K_M · 424.7 GB — Won't fit
Qwen3.5 397B-A17B
403B MoE · similar size
Q4_K_M · 227.4 GB — Won't fit
Ornith 1.5 397B-A17B
397B MoE · similar size
Q4_K_M · 224.1 GB — Won't fit

Every model against the 2× GeForce RTX 4090 →

DeepSeek V4.1 Flash 552B-A16B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
311.0 GB — Won't fit
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
311.0 GB — Won't fit
M4 · 16 GB
10.7 GB usable · 120 GB/s
311.0 GB — Won't fit
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
311.0 GB — Won't fit
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
311.0 GB — Won't fit
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
311.0 GB — Won't fit
M3 Ultra · 512 GB
384.0 GB usable · 800 GB/s
311.0 GB 23 Runs great

All 114 devices, by tier →