Deepseek-V3.2 needs about 407.0 GB of memory at Q4_K_M (a 405.4 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 27 quants, from 161.3 GB (TQ1_0) to 1342.3 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.
Deepseek‑V3.2 – 670 Billion‑parameter LLM with efficient long‑context reasoning
Deepseek‑V3.2 is a 670 Billion‑parameter model built on the DeepSeek‑2 architecture. It supports up to 163,840‑token context and is released under the MIT license. The model is optimized for reasoning and agentic tasks with a sparse attention mechanism.
Good at
- Long‑context reasoning
- Agentic tool use
- Competitive performance on math and coding contests
Limits
- Requires large VRAM for full precision
- No vision or multimodal support
- No built‑in chat template
Best for
- Developers building reasoning‑heavy agents
- Researchers testing long‑context performance
- Educational tools needing large‑context language models
Who it’s for. Engineers and researchers who need a high‑capacity language model for long‑context reasoning and agentic applications. The model is suitable for environments where GPU memory is sufficient to load the chosen quantization.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 161.3 GB | 162.9 GB | Too big | Get file → | |
| IQ1_S | 184.1 GB (4 parts) | 185.7 GB | Too big | Get file → | |
| IQ1_M | 199.3 GB (5 parts) | 200.9 GB | Too big | Get file → | |
| IQ2_XXS | 217.1 GB (5 parts) | 218.7 GB | Too big | Get file → | |
| IQ2_M | 228.1 GB (5 parts) | 229.7 GB | Too big | Get file → | |
| Q2_K | 245.4 GB (5 parts) | 247.0 GB | Too big | Get file → | |
| Q2_K_L | 245.6 GB (5 parts) | 247.2 GB | Too big | Get file → | |
| Q2_K_XL | 246.6 GB (6 parts) | 248.2 GB | Too big | Get file → | |
| IQ3_XXS | 273.1 GB (6 parts) | 274.7 GB | Too big | Get file → | |
| Q3_K_S | 290.4 GB (6 parts) | 292.0 GB | Too big | Get file → | |
| Q3_K_M | 320.2 GB (7 parts) | 321.8 GB | Too big | Get file → | |
| Q3_K_XL | 321.0 GB (7 parts) | 322.6 GB | Too big | Get file → | |
| IQ4_XS | 358.3 GB (8 parts) | 359.9 GB | Too big | Get file → | |
| IQ4_NL | 379.1 GB (8 parts) | 380.7 GB | Too big | Get file → | |
| Q4_0 | 380.1 GB (8 parts) | 381.7 GB | Too big | Get file → | |
| Q4_K_S | 381.1 GB (8 parts) | 382.6 GB | Too big | Get file → | |
| Q4_K_M | 405.4 GB (9 parts) | 407.0 GB | Too big | Get file → | |
| Q4_K_XL | 407.8 GB (9 parts) | 409.4 GB | Too big | Get file → | |
| Q4_1 | 420.8 GB (9 parts) | 422.4 GB | Too big | Get file → | |
| Q5_K_S | 462.6 GB (10 parts) | 464.2 GB | Too big | Get file → | |
| Q5_K_M | 476.2 GB (10 parts) | 477.7 GB | Too big | Get file → | |
| Q5_K_XL | 481.9 GB (10 parts) | 483.5 GB | Too big | Get file → | |
| Q6_K | 551.3 GB (12 parts) | 552.9 GB | Too big | Get file → | |
| Q6_K_XL | 574.0 GB (12 parts) | 575.6 GB | Too big | Get file → | |
| Q8_0 | 713.3 GB (15 parts) | 714.9 GB | Too big | Get file → | |
| Q8_K_XL | 780.2 GB (17 parts) | 781.8 GB | Too big | Get file → | |
| BF16 | 1342.3 GB (30 parts) | 1343.9 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S (≈382 GB file, ≈383 GB memory at 8k context). For 16 GB VRAM, Q4_K_M (≈407 GB file, ≈407 GB memory) is appropriate. For 24 GB VRAM, Q5_K_S (≈463 GB file, ≈464 GB memory) offers a balance of size and performance. Memory estimates are approximate.
What GPU do you need for Deepseek-V3.2?
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-V3.2 quant should you choose?
Q4_K_M is the usual starting point for Deepseek-V3.2: it balances file size and quality well. Here it is a 405.4 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 TQ1_0 at 161.3 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Deepseek-V3.2
How much VRAM does Deepseek-V3.2 need?
About 407.0 GB at Q4_K_M with an 8K context window. That is the 405.4 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-V3.2 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 162.9 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Deepseek-V3.2 GGUF?
It depends on the quant: TQ1_0 is 161.3 GB, IQ1_S is 184.1 GB, IQ1_M is 199.3 GB, IQ2_XXS is 217.1 GB, IQ2_M is 228.1 GB, Q2_K is 245.4 GB and 21 more.
What context length does Deepseek-V3.2 support?
The GGUF header lists a trained context length of 163,840 tokens (160K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.
Where can I download Deepseek-V3.2 GGUF?
The files are on Hugging Face in the unsloth/DeepSeek-V3.2-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-V3.2 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‑V3.2 on 8 GB VRAM?
Use the Q4_K_S quantization; it requires about 383 GB of VRAM at 8k context, which is beyond 8 GB, so it is not feasible.
Does Deepseek‑V3.2 support vision or multimodal inputs?
No, the model does not include vision or multimodal capabilities.