Raon 1 mini preview needs about 38.0 GB of memory at Q4_K_M (a 19.7 GB file) with an 8K context window, so it fits on a 48 GB GPU. This repository has 30 quants, from 8.7 GB (Q2_0) to 57.6 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 71.3 GB.
Raon 1 mini preview – 28.8B parameter LLM with reasoning and tools support
Raon 1 mini preview is a 28.8B parameter model built on the raon1 architecture. It supports reasoning and tool usage. The model is available in 30 GGUF quantizations ranging from Q2_0 to BF16.
Good at
- Reasoning tasks
- Tool integration
- Long‑context handling
Limits
- No vision support
- Requires krafton‑ai/llama.cpp build
- Large memory for higher quantizations
Best for
- Chatbots needing reasoning
- Tool‑enabled assistants
- Long‑form content generation
Who it’s for. Developers building AI assistants that require reasoning and tool usage. The model is suitable for users who can run custom llama.cpp builds and have sufficient memory for the chosen quantization.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_0 | 8.7 GB | 27.1 GB | Too big | Get file → | |
| IQ1_S | 9.8 GB | 28.2 GB | Too big | Get file → | |
| IQ1_M | 10.3 GB | 28.6 GB | Too big | Get file → | |
| IQ2_XXS | 11.0 GB | 29.3 GB | Too big | Get file → | |
| IQ2_XS | 11.5 GB | 29.8 GB | Too big | Get file → | |
| IQ2_S | 11.6 GB | 29.9 GB | Too big | Get file → | |
| IQ2_M | 12.1 GB | 30.5 GB | Too big | Get file → | |
| Q2_K | 12.3 GB | 30.6 GB | Too big | Get file → | |
| Q2_K_S | 12.3 GB | 30.7 GB | Too big | Get file → | |
| IQ3_XXS | 13.2 GB | 31.5 GB | Too big | Get file → | |
| IQ3_XS | 13.5 GB | 31.9 GB | Too big | Get file → | |
| IQ3_S | 14.2 GB | 32.5 GB | Too big | Get file → | |
| Q3_K_S | 14.2 GB | 32.5 GB | Too big | Get file → | |
| IQ3_M | 14.3 GB | 32.6 GB | Too big | Get file → | |
| Q3_K_M | 15.3 GB | 33.6 GB | Too big | Get file → | |
| Q3_K_L | 15.8 GB | 34.2 GB | Too big | Get file → | |
| IQ4_XS | 16.1 GB | 34.4 GB | Too big | Get file → | |
| Q4_0 | 16.4 GB | 34.8 GB | Too big | Get file → | |
| IQ4_NL | 16.7 GB | 35.0 GB | Too big | Get file → | |
| MXFP4_MOE | 17.0 GB | 35.3 GB | Too big | Get file → | |
| Q4_K_S | 18.1 GB | 36.4 GB | Too big | Get file → | |
| Q4_1 | 18.2 GB | 36.5 GB | Too big | Get file → | |
| Q4_K_M | 19.7 GB | 38.0 GB | Too big | Get file → | |
| Q5_0 | 20.0 GB | 38.3 GB | Too big | Get file → | |
| Q5_K_S | 20.8 GB | 39.2 GB | Too big | Get file → | |
| Q5_1 | 21.7 GB | 40.0 GB | Too big | Get file → | |
| Q5_K_M | 22.2 GB | 40.5 GB | Too big | Get file → | |
| Q6_K | 25.8 GB | 44.2 GB | Too big | Get file → | |
| Q8_0 | 30.7 GB | 49.0 GB | Too big | Get file → | |
| BF16 | 57.6 GB | 75.9 GB | Too big | Get file → |
For 8 GB VRAM, use IQ4_XS (file size 16.07 GB, memory 34.4 GB at 8k context). For 16 GB VRAM, Q4_K_M (file size 19.69 GB, memory 38 GB) is a good balance. For 32 GB VRAM, Q6_K (file size 25.84 GB, memory 44.2 GB) offers higher quality.
What GPU do you need for Raon 1 mini preview?
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: the best quant that fits is IQ3_XS (13.5 GB file, about 31.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q6_K (25.8 GB file, about 44.2 GB in use).
Which Raon 1 mini preview quant should you choose?
Q4_K_M is the usual starting point for Raon 1 mini preview: it balances file size and quality well. Here it is a 19.7 GB file.
The Q8_0 file is 30.7 GB, about 1.6 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 Q2_0 at 8.7 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 Raon 1 mini preview
How much VRAM does Raon 1 mini preview need?
About 38.0 GB at Q4_K_M with an 8K context window. That is the 19.7 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 Raon 1 mini preview on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 27.1 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Raon 1 mini preview GGUF?
It depends on the quant: Q2_0 is 8.7 GB, IQ1_S is 9.8 GB, IQ1_M is 10.3 GB, IQ2_XXS is 11.0 GB, IQ2_XS is 11.5 GB, IQ2_S is 11.6 GB and 24 more.
What context length does Raon 1 mini preview support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 17.8 GB at 8K and 71.3 GB at 32K.
Where can I download Raon 1 mini preview GGUF?
The files are on Hugging Face in the NANI-Nithin/Raon-1-mini-preview-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 Raon 1 mini preview 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.
Can I run Raon 1 mini preview on 8 GB VRAM?
Use the IQ4_XS quantization; it requires about 34.4 GB of RAM at 8k context, so 8 GB VRAM is insufficient for inference.
What build of llama.cpp is required?
The model needs krafton‑ai/llama.cpp at the raon‑1‑mini‑preview branch; it is not yet in upstream llama.cpp.
Does the model support vision?
No, the model has no vision capabilities.