DE

DeepSeek V3 0324

DeepSeekGGUF by bartowskiUpdated 2025-04-04
Download on Hugging Face
Parameters670B
Context160K
Layers61
Architecturedeepseek2
Smallest quant133.6 GB
Quick answer

DeepSeek V3 0324 needs about 445.9 GB of memory at Q4_K_M (a 404.4 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 26 quants, from 133.6 GB (IQ1_S) to 713.3 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 163.7 GB.

DeepSeek V3 0324 – 670B MoE model with extensive quantization options

DeepSeek V3 0324 is a 670 billion‑parameter mixture‑of‑experts model. It supports function calling, JSON output, and FIM completion. The model is released under the MIT license and is available in multiple GGUF quantizations.

Good at

  • Strong reasoning and math skills
  • Web development code generation
  • Chinese writing and translation

Limits

  • Requires large VRAM for higher‑quality quants
  • No vision support
  • Context limited to 163 840 tokens

Best for

  • Complex code generation
  • Long‑form Chinese content creation
  • Multi‑turn interactive rewriting

Who it’s for. Developers and researchers needing a large MoE model for advanced text generation. The model is suitable for local deployment with llama.cpp or compatible frameworks.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S133.6 GB (4 parts) 175.0 GB
Too big Get file →
IQ1_M148.9 GB (4 parts) 190.3 GB
Too big Get file →
IQ1_M154.8 GB (4 parts) 196.2 GB
Too big Get file →
IQ2_XXS174.4 GB (5 parts) 215.9 GB
Too big Get file →
IQ2_XXS179.8 GB (5 parts) 221.2 GB
Too big Get file →
IQ2_S197.0 GB (6 parts) 238.4 GB
Too big Get file →
IQ2_M217.4 GB (6 parts) 258.9 GB
Too big Get file →
IQ2_M224.5 GB (6 parts) 265.9 GB
Too big Get file →
Q2_K_L244.9 GB (7 parts) 286.4 GB
Too big Get file →
Q2_K_L247.3 GB (7 parts) 288.8 GB
Too big Get file →
Q2_K248.9 GB (7 parts) 290.3 GB
Too big Get file →
IQ3_XXS257.9 GB (7 parts) 299.4 GB
Too big Get file →
IQ3_XXS261.7 GB (7 parts) 303.2 GB
Too big Get file →
Q3_K_M319.2 GB (9 parts) 360.6 GB
Too big Get file →
Q3_K_XL348.3 GB (9 parts) 389.7 GB
Too big Get file →
IQ4_XS357.1 GB (10 parts) 398.6 GB
Too big Get file →
IQ4_NL378.1 GB (10 parts) 419.5 GB
Too big Get file →
Q4_0379.0 GB (10 parts) 420.5 GB
Too big Get file →
Q4_K_S380.0 GB (10 parts) 421.4 GB
Too big Get file →
Q4_K_M404.4 GB (11 parts) 445.9 GB
Too big Get file →
Q4_K_M407.0 GB (11 parts) 448.4 GB
Too big Get file →
Q4_1419.9 GB (11 parts) 461.4 GB
Too big Get file →
Q5_K_S461.8 GB (12 parts) 503.3 GB
Too big Get file →
Q5_K_M475.4 GB (13 parts) 516.8 GB
Too big Get file →
Q6_K550.8 GB (15 parts) 592.2 GB
Too big Get file →
Q8_0713.3 GB (20 parts) 754.7 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (file size 133.56 GB, memory 175 GB at 8 k context). For 16 GB VRAM, IQ2_S (file size 197 GB, memory 238.4 GB). For 24 GB VRAM, Q4_K_M (file size 404.43 GB, memory 445.9 GB). Estimates are approximate.

What GPU do you need for DeepSeek V3 0324?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: no quant fits at an 8K context.
  • 32 GB VRAM: no quant fits at an 8K context.
  • 48 GB VRAM: no quant fits at an 8K context.

Which DeepSeek V3 0324 quant should you choose?

Q4_K_M is the usual starting point for DeepSeek V3 0324: it balances file size and quality well. Here it is a 404.4 GB file.

The Q8_0 file is 713.3 GB, about 1.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is IQ1_S at 133.6 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturedeepseek2
Parameters670B (approximate, mixture of experts)
Layers61
Trained context163,840 tokens
KV cache (fp16)4880 KB per token

Source

Quantized bybartowski
FamilyDeepSeek
Licensemit
Downloads (HF)1,825
Last updated2025-04-04

Frequently asked questions about DeepSeek V3 0324

How much VRAM does DeepSeek V3 0324 need?

About 445.9 GB at Q4_K_M with an 8K context window. That is the 404.4 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run DeepSeek V3 0324 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 175.0 GB, so you would need a larger GPU or CPU offloading.

What is the file size of DeepSeek V3 0324 GGUF?

It depends on the quant: IQ1_S is 133.6 GB, IQ1_M is 148.9 GB, IQ1_M is 154.8 GB, IQ2_XXS is 174.4 GB, IQ2_XXS is 179.8 GB, IQ2_S is 197.0 GB and 20 more.

What context length does DeepSeek V3 0324 support?

The GGUF header lists a trained context length of 163,840 tokens (160K). The KV cache grows with the context you use: about 40.9 GB at 8K and 163.7 GB at 32K.

Where can I download DeepSeek V3 0324 GGUF?

The files are on Hugging Face in the bartowski/deepseek-ai_DeepSeek-V3-0324-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does DeepSeek V3 0324 use?

Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run DeepSeek V3 0324 on 8 GB VRAM?

Use the IQ1_S quantization; it requires about 175 GB of memory at 8 k context, which is beyond 8 GB VRAM, so it is not feasible.

Does the model support vision tasks?

No, the model does not have vision capabilities.

What is the maximum context length?

The trained context length is 163 840 tokens.

Other GGUFs of DeepSeek-V3-0324

Similar models

Browse all →

More from bartowski

All by bartowski →

More DeepSeek models

All DeepSeek →

Other large models

All large models →