granite-guardian-3.2-3b-a800m needs about 2.9 GB of memory at IQ4_NL (a 1.9 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 21 quants, from 1.2 GB (Q2_K_S) to 6.6 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 2.1 GB.
granite-guardian-3.2-3b-a800m is a 3.3B-parameter Granite model in GGUF format, quantized by Mungert. Its trained context length is 128K tokens. The repository has 21 quant files, from 1.2 GB (Q2_K_S) to 6.6 GB (BF16).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K_S | 1.2 GB | 2.2 GB | Fits | Get file → | |
| Q2_K_M | 1.3 GB | 2.3 GB | Fits | Get file → | |
| IQ3_XXS | 1.3 GB | 2.4 GB | Fits | Get file → | |
| IQ3_XS | 1.4 GB | 2.4 GB | Fits | Get file → | |
| IQ3_M | 1.5 GB | 2.5 GB | Fits | Get file → | |
| IQ3_S | 1.5 GB | 2.5 GB | Fits | Get file → | |
| Q3_K_S | 1.5 GB | 2.6 GB | Fits | Get file → | |
| Q3_K_M | 1.6 GB | 2.7 GB | Fits | Get file → | |
| IQ4_XS | 1.8 GB | 2.8 GB | Fits | Get file → | |
| Q4_0 | 1.9 GB | 2.9 GB | Fits | Get file → | |
| IQ4_NL | 1.9 GB | 2.9 GB | Fits | Get file → | |
| Q4_1 | 2.1 GB | 3.1 GB | Fits | Get file → | |
| Q5_0 | 2.3 GB | 3.3 GB | Fits | Get file → | |
| Q5_K_S | 2.4 GB | 3.4 GB | Fits | Get file → | |
| Q5_K_M | 2.4 GB | 3.5 GB | Fits | Get file → | |
| Q5_1 | 2.5 GB | 3.5 GB | Fits | Get file → | |
| Q6_K_M | 2.7 GB | 3.8 GB | Fits | Get file → | |
| Q8_0 | 3.5 GB | 4.5 GB | Fits | Get file → | |
| BF16 | 3.7 GB | 4.7 GB | Fits | Get file → | |
| F16 | 3.7 GB | 4.7 GB | Fits | Get file → | |
| BF16 | 6.6 GB | 7.6 GB | Tight | Get file → |
What GPU do you need for granite-guardian-3.2-3b-a800m?
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: the best quant that fits is Q6_K_M (2.7 GB file, about 3.8 GB in use).
- 6 GB VRAM: the best quant that fits is BF16 (3.7 GB file, about 4.7 GB in use).
- 8 GB VRAM: the best quant that fits is BF16 (6.6 GB file, about 7.6 GB in use).
Which granite-guardian-3.2-3b-a800m quant should you choose?
IQ4_NL (1.9 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is Q2_K_S at 1.2 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
Frequently asked questions about granite-guardian-3.2-3b-a800m
How much VRAM does granite-guardian-3.2-3b-a800m need?
About 2.9 GB at IQ4_NL with an 8K context window. That is the 1.9 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 granite-guardian-3.2-3b-a800m on an 8 GB GPU?
Yes, with BF16: the 6.6 GB file needs about 7.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of granite-guardian-3.2-3b-a800m GGUF?
It depends on the quant: Q2_K_S is 1.2 GB, Q2_K_M is 1.3 GB, IQ3_XXS is 1.3 GB, IQ3_XS is 1.4 GB, IQ3_M is 1.5 GB, IQ3_S is 1.5 GB and 15 more.
What context length does granite-guardian-3.2-3b-a800m support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 2.1 GB at 32K.
Where can I download granite-guardian-3.2-3b-a800m GGUF?
The files are on Hugging Face in the Mungert/granite-guardian-3.2-3b-a800m-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 granite-guardian-3.2-3b-a800m use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.