Carbon 500M needs about 1.4 GB of memory at Q4_K_M (a 0.4 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 5 quants, from 0.4 GB (Q4_K_M) to 1.0 GB (BF16).
Carbon 500M – a 512‑million‑parameter DNA generative model
Carbon 500M is a 512‑million‑parameter decoder‑only autoregressive model built on a Llama‑style architecture. It uses a hybrid tokenizer that combines 6‑mer DNA tokens with Qwen3 BPE for English text. The model supports a native context length of 8,192 tokens (≈49 kbp).
Good at
- DNA sequence generation
- Hybrid DNA‑text tokenization
- Speculative decoding with larger Carbon models
Limits
- No long‑context training beyond 8,192 tokens
- Not a chat model, only text completion
- License is apache‑2.0
Best for
- Drafting DNA sequences for larger models
- Generating hybrid DNA‑text prompts
- Speculative decoding to reduce generation cost
Who it’s for. Researchers and developers working with DNA generative models who need a lightweight base model for drafting or speculative decoding. It is suitable for those who prefer a Llama‑style architecture with a hybrid tokenizer.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 0.4 GB | 1.4 GB | Fits | Get file → | |
| Q5_K_M | 0.4 GB | 1.4 GB | Fits | Get file → | |
| Q6_K | 0.5 GB | 1.4 GB | Fits | Get file → | |
| Q8_0 | 0.6 GB | 1.5 GB | Fits | Get file → | |
| BF16 | 1.0 GB | 2.0 GB | Fits | Get file → |
For 8 GB VRAM, use Q4_K_M or Q5_K_M (0.39 GB or 0.42 GB). For 16 GB VRAM, Q6_K (0.46 GB) or Q8_0 (0.55 GB) are suitable. If more memory is available, BF16 (1.03 GB) provides higher precision. Memory estimates are approximate.
What GPU do you need for Carbon 500M?
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 BF16 (1.0 GB file, about 2.0 GB in use).
Which Carbon 500M quant should you choose?
Q4_K_M is the usual starting point for Carbon 500M: it balances file size and quality well. Here it is a 0.4 GB file.
The Q8_0 file is 0.6 GB, about 1.4 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 500M
How much VRAM does Carbon 500M need?
About 1.4 GB at Q4_K_M with an 8K context window. That is the 0.4 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 500M on an 8 GB GPU?
Yes, with BF16: the 1.0 GB file needs about 2.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Carbon 500M GGUF?
It depends on the quant: Q4_K_M is 0.4 GB, Q5_K_M is 0.4 GB, Q6_K is 0.5 GB, Q8_0 is 0.6 GB, BF16 is 1.0 GB.
What context length does Carbon 500M support?
The GGUF header lists a trained context length of 8,192 tokens (8K). The KV cache grows with the context you use: about 0.5 GB at 8K and 0.5 GB at 8K.
Where can I download Carbon 500M GGUF?
The files are on Hugging Face in the HuggingFaceBio/Carbon-500M-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 500M 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 500M on 8 GB VRAM?
Yes, the Q4_K_M and Q5_K_M quantizations are under 0.5 GB and require about 1.4 GB of VRAM at 8,192‑token context.
Does Carbon 500M support vision tasks?
No, it is a text‑only DNA generative model with no vision capabilities.
Is Carbon 500M suitable for chat applications?
No, it is a base DNA model intended for text completion, not chat.