grok 2 needs about 166.7 GB of memory at Q4_K_M (a 164.1 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 27 quants, from 81.8 GB (TQ1_0) to 539.1 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
Grok 2 – a 270 B parameter Mixture‑of‑Experts model with 131 072‑token context
Grok 2 is a 270 B parameter model trained by xAI in 2024. It uses a Mixture‑of‑Experts architecture and supports a 131 072‑token context window. The weights are available in GGUF format and are licensed under the Grok 2 Community License Agreement.
Good at
- Large context handling
- Mixture‑of‑Experts inference
- High‑capacity language modeling
Limits
- Requires 8 GPUs with >40 GB VRAM for full precision
- No vision support
- License restricts commercial use
Best for
- Long‑form content generation
- Complex reasoning tasks
- Research prototyping
Who it’s for. Developers and researchers needing a large‑context, Mixture‑of‑Experts model for text‑only workloads. The model is suitable for environments with multiple high‑memory GPUs.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 81.8 GB | 84.4 GB | Too big | Get file → | |
| IQ1_S | 88.9 GB (2 parts) | 91.6 GB | Too big | Get file → | |
| IQ1_M | 94.5 GB (2 parts) | 97.2 GB | Too big | Get file → | |
| IQ2_XXS | 99.3 GB (2 parts) | 102.0 GB | Too big | Get file → | |
| Q2_K | 100.1 GB (3 parts) | 102.7 GB | Too big | Get file → | |
| Q2_K_L | 100.3 GB (3 parts) | 102.9 GB | Too big | Get file → | |
| IQ2_M | 103.1 GB (3 parts) | 105.8 GB | Too big | Get file → | |
| Q2_K_XL | 112.1 GB (3 parts) | 114.7 GB | Too big | Get file → | |
| IQ3_XXS | 117.3 GB (3 parts) | 120.0 GB | Too big | Get file → | |
| Q3_K_S | 118.0 GB (3 parts) | 120.7 GB | Too big | Get file → | |
| Q3_K_XL | 126.5 GB (3 parts) | 129.1 GB | Too big | Get file → | |
| Q3_K_M | 130.1 GB (3 parts) | 132.8 GB | Too big | Get file → | |
| IQ4_XS | 144.8 GB (3 parts) | 147.4 GB | Too big | Get file → | |
| IQ4_NL | 153.0 GB (4 parts) | 155.6 GB | Too big | Get file → | |
| Q4_0 | 153.1 GB (4 parts) | 155.8 GB | Too big | Get file → | |
| Q4_K_S | 154.3 GB (4 parts) | 157.0 GB | Too big | Get file → | |
| Q4_K_XL | 154.7 GB (4 parts) | 157.4 GB | Too big | Get file → | |
| Q4_K_M | 164.1 GB (4 parts) | 166.7 GB | Too big | Get file → | |
| Q4_1 | 169.2 GB (4 parts) | 171.8 GB | Too big | Get file → | |
| Q5_K_S | 185.9 GB (4 parts) | 188.5 GB | Too big | Get file → | |
| Q5_K_XL | 191.0 GB (4 parts) | 193.7 GB | Too big | Get file → | |
| Q5_K_M | 191.6 GB (4 parts) | 194.2 GB | Too big | Get file → | |
| Q6_K | 221.4 GB (5 parts) | 224.0 GB | Too big | Get file → | |
| Q6_K_XL | 232.0 GB (5 parts) | 234.7 GB | Too big | Get file → | |
| Q8_0 | 286.4 GB (6 parts) | 289.0 GB | Too big | Get file → | |
| Q8_K_XL | 315.8 GB (7 parts) | 318.5 GB | Too big | Get file → | |
| BF16 | 539.1 GB (11 parts) | 541.7 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S or Q4_K_XL (≈157 GB memory estimate). For 16 GB VRAM, Q5_K_S or Q5_K_M (≈188–194 GB). For 32 GB VRAM, Q6_K or Q6_K_XL (≈224–235 GB). Estimates are approximate.
What GPU do you need for grok 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 grok 2 quant should you choose?
Q4_K_M is the usual starting point for grok 2: it balances file size and quality well. Here it is a 164.1 GB file.
The Q8_0 file is 286.4 GB, about 1.7 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 81.8 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 grok 2
How much VRAM does grok 2 need?
About 166.7 GB at Q4_K_M with an 8K context window. That is the 164.1 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 grok 2 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 84.4 GB, so you would need a larger GPU or CPU offloading.
What is the file size of grok 2 GGUF?
It depends on the quant: TQ1_0 is 81.8 GB, IQ1_S is 88.9 GB, IQ1_M is 94.5 GB, IQ2_XXS is 99.3 GB, Q2_K is 100.1 GB, Q2_K_L is 100.3 GB and 21 more.
What context length does grok 2 support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.6 GB at 32K.
Where can I download grok 2 GGUF?
The files are on Hugging Face in the unsloth/grok-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 grok 2 use?
Hugging Face lists the license as other. 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 Grok 2 on a single 8 GB GPU?
No, the model requires at least 8 GPUs with >40 GB VRAM each for full precision. Quantized versions still need large VRAM.
Does Grok 2 support image input?
No, the model is text‑only and has no vision capabilities.
What license applies to Grok 2 weights?
The weights are licensed under the Grok 2 Community License Agreement, which may restrict commercial use.