Nanbeige4.3 33B UNCENSORED needs about 22.3 GB of memory at Q4_K_M (a 20.1 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 3 quants, from 9.3 GB (IQ2_XXS) to 27.2 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 7.0 GB.
Nanbeige4.3 33B UNCENSORED – a 33.1B parameter Llama model for agentic reasoning
Nanbeige4.3 33B UNCENSORED is a 33.1B parameter Llama-family model with 52 layers and 7168 hidden size. It supports a 512K token context and includes native think on/off mode. The model is released in three quantizations: IQ2_XXS, Q4_K_M, and Q6_K.
Good at
- Reasoning
- Tools usage
- Agentic coding
Limits
- 512K context only
- No vision support
- License is other
Best for
- Complex coding tasks
- Professional coworking
- Agentic automation
Who it’s for. Developers and researchers needing a large Llama model for agentic and reasoning workloads. The model is suitable for experimentation but requires downstream safety controls before end‑user deployment.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_XXS | 9.3 GB | 11.6 GB | Too big | Get file → | |
| Q4_K_M | 20.1 GB | 22.3 GB | Too big | Get file → | |
| Q6_K | 27.2 GB | 29.4 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (9.33 GB, ~11.6 GB memory at 8 k context). For 16 GB VRAM, Q4_K_M (20.05 GB, ~22.3 GB memory). For 24 GB VRAM, Q6_K (27.17 GB, ~29.4 GB memory). Estimates are approximate.
What GPU do you need for Nanbeige4.3 33B UNCENSORED?
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: the best quant that fits is IQ2_XXS (9.3 GB file, about 11.6 GB in use).
- 16 GB VRAM: the best quant that fits is IQ2_XXS (9.3 GB file, about 11.6 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_K_M (20.1 GB file, about 22.3 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (27.2 GB file, about 29.4 GB in use).
Which Nanbeige4.3 33B UNCENSORED quant should you choose?
Q4_K_M is the usual starting point for Nanbeige4.3 33B UNCENSORED: it balances file size and quality well. Here it is a 20.1 GB file.
If memory is tight, the smallest file is IQ2_XXS at 9.3 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
Source
Frequently asked questions about Nanbeige4.3 33B UNCENSORED
How much VRAM does Nanbeige4.3 33B UNCENSORED need?
About 22.3 GB at Q4_K_M with an 8K context window. That is the 20.1 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 Nanbeige4.3 33B UNCENSORED on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 11.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Nanbeige4.3 33B UNCENSORED GGUF?
It depends on the quant: IQ2_XXS is 9.3 GB, Q4_K_M is 20.1 GB, Q6_K is 27.2 GB.
What context length does Nanbeige4.3 33B UNCENSORED support?
The GGUF header lists a trained context length of 524,288 tokens (512K). The KV cache grows with the context you use: about 1.7 GB at 8K and 7.0 GB at 32K.
Where can I download Nanbeige4.3 33B UNCENSORED GGUF?
The files are on Hugging Face in the dealignai/Nanbeige4.3-33B-UNCENSORED-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 Nanbeige4.3 33B UNCENSORED 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 Nanbeige4.3 33B UNCENSORED on 8 GB VRAM?
Only the IQ2_XXS quant (~9.33 GB) may fit, but memory usage will be ~11.6 GB at 8 k context, so 8 GB VRAM is insufficient.
Does this model support vision?
No, the model has no vision capability.
What license applies to this model?
The model is released under an "other" license; consult the repository for details.