K2-Chat needs about 50.1 GB of memory at Q3_K_S (a 28.2 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 24 quants, from 14.2 GB (IQ1_S) to 130.6 GB.
K2-Chat is a 65.3B-parameter Chat model in GGUF format, quantized by legraphista. Its trained context length is 8K tokens. The repository has 24 quant files, from 14.2 GB (IQ1_S) to 130.6 GB.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ1_S | 14.2 GB | 36.2 GB | Too big | Get file → | |
| IQ1_M | 15.4 GB | 37.4 GB | Too big | Get file → | |
| IQ2_XXS | 17.5 GB | 39.4 GB | Too big | Get file → | |
| IQ2_XS | 19.3 GB | 41.2 GB | Too big | Get file → | |
| IQ2_S | 20.8 GB | 42.8 GB | Too big | Get file → | |
| Q2_K_S | 22.0 GB | 44.0 GB | Too big | Get file → | |
| IQ2_M | 22.4 GB | 44.4 GB | Too big | Get file → | |
| Q2_K | 24.1 GB | 46.1 GB | Too big | Get file → | |
| IQ3_XXS | 24.7 GB | 46.6 GB | Too big | Get file → | |
| IQ3_XS | 26.6 GB | 48.6 GB | Too big | Get file → | |
| IQ3_S | 28.2 GB | 50.1 GB | Too big | Get file → | |
| Q3_K_S | 28.2 GB | 50.1 GB | Too big | Get file → | |
| IQ3_M | 29.8 GB | 51.8 GB | Too big | Get file → | |
| Q3_K | 31.6 GB | 53.6 GB | Too big | Get file → | |
| Q3_K_L | 34.6 GB | 56.6 GB | Too big | Get file → | |
| IQ4_XS | 34.8 GB | 56.7 GB | Too big | Get file → | |
| IQ4_NL | 36.8 GB | 58.8 GB | Too big | Get file → | |
| Q4_K_S | 37.1 GB | 59.0 GB | Too big | Get file → | |
| Q4_K | 39.3 GB | 61.3 GB | Too big | Get file → | |
| Q5_K_S | 44.9 GB | 66.9 GB | Too big | Get file → | |
| Q5_K | 46.2 GB (2 parts) | 68.2 GB | Too big | Get file → | |
| Q6_K | 53.6 GB (3 parts) | 75.5 GB | Too big | Get file → | |
| Q8_0 | 69.4 GB (3 parts) | 91.3 GB | Too big | Get file → | |
| — | 130.6 GB (6 parts) | 152.6 GB | Too big | Get file → |
What GPU do you need for K2-Chat?
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: the best quant that fits is IQ3_XXS (24.7 GB file, about 46.6 GB in use).
Which K2-Chat quant should you choose?
Q3_K_S (28.2 GB) is the middle option in this repository. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
If memory is tight, the smallest file is IQ1_S at 14.2 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
Source
Frequently asked questions about K2-Chat
How much VRAM does K2-Chat need?
About 50.1 GB at Q3_K_S with an 8K context window. That is the 28.2 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 K2-Chat on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 36.2 GB, so you would need a larger GPU or CPU offloading.
What is the file size of K2-Chat GGUF?
It depends on the quant: IQ1_S is 14.2 GB, IQ1_M is 15.4 GB, IQ2_XXS is 17.5 GB, IQ2_XS is 19.3 GB, IQ2_S is 20.8 GB, Q2_K_S is 22.0 GB and 18 more.
What context length does K2-Chat support?
The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 21.5 GB at 8K and 21.5 GB at 8K.
Where can I download K2-Chat GGUF?
The files are on Hugging Face in the legraphista/K2-Chat-IMat-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 K2-Chat 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.