Carbon-X needs about 11.5 GB of memory at Q4_K_M (a 9.6 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 5 quants, from 9.6 GB (Q4_K_M) to 29.7 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.
Carbon‑X: 14.8B DNA‑aware language model
Carbon‑X is a 14.8B‑parameter decoder‑only model trained on DNA sequences and natural‑language text. It predicts mutation effects, identifies regulatory regions, classifies organisms, and generates next bases while explaining its reasoning.
Good at
- DNA mutation prediction
- Regulatory region identification
- Next‑base generation
Limits
- No vision support
- Requires llama.cpp b11742 or newer
- License is apache‑2.0
Best for
- Variant effect analysis
- Gene regulation studies
- DNA sequence generation
Who it’s for. Researchers and developers needing local inference of DNA‑related tasks. The model can run on private infrastructure, keeping sensitive sequences secure.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 9.6 GB | 11.5 GB | Too big | Get file → | |
| Q5_K_M | 11.1 GB | 12.9 GB | Too big | Get file → | |
| Q6_K | 12.6 GB | 14.4 GB | Too big | Get file → | |
| Q8_0 | 15.8 GB | 17.6 GB | Too big | Get file → | |
| BF16 | 29.7 GB | 31.5 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (9.64 GB, ~11.5 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (11.05 GB, ~12.9 GB memory). For 16 GB VRAM, Q6_K (12.56 GB, ~14.4 GB memory). The Q8_0 file is recommended for best performance if memory allows.
What GPU do you need for Carbon-X?
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 Q4_K_M (9.6 GB file, about 11.5 GB in use).
- 16 GB VRAM: the best quant that fits is Q6_K (12.6 GB file, about 14.4 GB in use).
- 24 GB VRAM: the best quant that fits is Q8_0 (15.8 GB file, about 17.6 GB in use).
- 32 GB VRAM: the best quant that fits is BF16 (29.7 GB file, about 31.5 GB in use).
Which Carbon-X quant should you choose?
Q4_K_M is the usual starting point for Carbon-X: it balances file size and quality well. Here it is a 9.6 GB file.
The Q8_0 file is 15.8 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.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Carbon-X
How much VRAM does Carbon-X need?
About 11.5 GB at Q4_K_M with an 8K context window. That is the 9.6 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 Carbon-X on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 11.5 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Carbon-X GGUF?
It depends on the quant: Q4_K_M is 9.6 GB, Q5_K_M is 11.1 GB, Q6_K is 12.6 GB, Q8_0 is 15.8 GB, BF16 is 29.7 GB.
What context length does Carbon-X support?
The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 1.3 GB at 8K and 5.4 GB at 32K.
Where can I download Carbon-X GGUF?
The files are on Hugging Face in the HuggingFaceBio/Carbon-X-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 Carbon-X 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 Carbon‑X on 8 GB VRAM?
Yes, using the Q4_K_M quantization which requires about 11.5 GB memory at 8 k context.
Does Carbon‑X support image input?
No, the model has no vision capabilities.
What license applies to Carbon‑X?
Carbon‑X is released under the apache‑2.0 license.