DeepSeek V3.1 (Q4_K_M) — 54.7 GBon NVIDIA L4

Q4_K_M NVIDIA L4

Overview

DeepSeek V3.1 is a 684.53B parameter moe language model by DeepSeek, with code, multilingual, thinking, tool-calls capabilities. It supports a context window of up to 163,840 tokens.

DeepSeek V3.1 is a 685-billion-parameter Mixture-of-Experts model from DeepSeek, activating 8 of 256 experts per token plus one shared expert. It delivers frontier-level performance on code generation, reasoning, and multilingual tasks while using far fewer active parameters per inference step than comparably sized dense models. The model supports thinking mode, tool calling, and nine languages. With a 160K context window, it requires multi-GPU or distributed setups but quantizes down to Q2 levels for reduced VRAM footprint.

At Q4_K_M quantization (medium quality tier), the model weighs 377.56 GB. This exceeds the 24 GB of VRAM on NVIDIA L4. Inference is still possible via CPU offload or memory-mapped loading from disk, but expect significantly reduced performance.

The NVIDIA L4 is a datacenter inference GPU with 24 GB of GDDR6 VRAM and 300 GB/s memory bandwidth. It delivers 121 FP16 TFLOPS with Ada Lovelace architecture. Designed for efficient, low-power inference workloads in cloud and edge deployments. Handles quantized models up to 20B parameters.

Hardware Requirements

Model size	377.56 GB
VRAM available	24 GB
VRAM used	54.7 GB
Min RAM required	377.6 GB
GPU layers	0 / 61
Context size	32,768
Backend	cuda13
Flash attention	No

Performance Notes

Mixture-of-Experts

This is a Mixture-of-Experts model with 256 total experts, of which 8 are active per token. Only the active experts need fast GPU access — inactive expert weights can reside in system RAM with minimal impact on generation speed.

Read the MoE deployment guide →

Reduced performance

61 of 61 layers run on CPU. Your system needs at least 377.6 GB RAM. If system RAM is insufficient, data will be read from disk on demand, with significant performance impact.

Shorter conversations

Context size has been reduced due to memory constraints. This may affect the model's ability to handle long conversations.

Deploy

Server Local

Prerequisites

Ensure your GPU nodes are prepared with the NVIDIA container toolkit:

ansible-playbook prositronic.infra.nvidia_container_toolkit

Command

helmfile --state-values-file <(curl -s https://www.prositronic.eu/values/deepseek-v3-1/q4_k_m/nvidia-l4.yaml) apply

Generated values.yaml

/values/deepseek-v3-1/q4_k_m/nvidia-l4.yaml

Loading values…

Install llama.cpp

Install llama.cpp from the official build instructions for your platform and backend.

Download Model

curl -L -o deepseek-v3-1.gguf "https://huggingface.co/unsloth/DeepSeek-V3.1-GGUF/resolve/main/Q4_K_M/DeepSeek-V3.1-Q4_K_M-00001-of-00009.gguf"

Start Server

llama-server \
  -m deepseek-v3-1.gguf \
  --n-gpu-layers 0 \
  --ctx-size 32768

Verify

curl http://localhost:8080/health

Explore

• Configure a deployment
• Browse models
• Browse hardware

Other quantizations

• DeepSeek V3.1 (Q8_0)
• DeepSeek V3.1 (Q5_K_M)

Other hardware

• DeepSeek V3.1 on CPU Only
• DeepSeek V3.1 on NVIDIA A100 80GB
• DeepSeek V3.1 on NVIDIA H100

Frequently Asked Questions

How much VRAM does DeepSeek V3.1 (Q4_K_M) need?

The Q4_K_M quantization of DeepSeek V3.1 requires 377.56 GB. The 24 GB of VRAM on NVIDIA L4 is insufficient for GPU layers, so inference runs on CPU.

Can I run DeepSeek V3.1 on NVIDIA L4?

It is possible but not recommended. NVIDIA L4 does not have enough VRAM to accelerate DeepSeek V3.1 (Q4_K_M), so inference will rely on CPU and system RAM.

What is quantization?

Quantization reduces a model's numerical precision from its original floating-point format to a more compact representation. This shrinks the file size and VRAM footprint, making it possible to run large models on consumer hardware. The trade-off is a small reduction in output quality. Q4_K_M compresses DeepSeek V3.1 from its original size down to 377.56 GB.

What quantization should I choose for DeepSeek V3.1?

Q4_K_M is a medium-quality quantization. Higher-quality quants (Q8, Q6) preserve more model accuracy but need more VRAM. Lower quants (Q4, Q3, Q2) reduce VRAM usage at the cost of some quality. Choose based on your available hardware and quality requirements.

Why are some layers offloaded to CPU?

NVIDIA L4 has 24 GB of VRAM, but DeepSeek V3.1 (Q4_K_M) requires approximately 377.56 GB. Only 0 of 61 layers fit in VRAM; the remaining layers run on CPU, which is slower but still functional.

What is MoE and how does it affect deployment?

DeepSeek V3.1 uses a Mixture-of-Experts (MoE) architecture with 256 experts, of which 8 are active per token. This means only a fraction of the model weights are used for each inference step, allowing MoE models to be larger in total parameter count while remaining efficient at inference time.

Last updated: March 5, 2026