Carbon-X
Carbon-X is an open 14.8B-parameter decoder-only language model trained jointly on DNA sequences and natural-language text. It accepts DNA in chat prompts and can respond in English, DNA, or both, with capabilities including variant-effect prediction, sequence understanding, and next-base prediction.
What it does
Researchers can use it to analyze DNA sequences, assess variant effects, identify sequence features, and predict subsequent DNA bases.
- Availability
- Model weights and usage instructions are provided on Hugging Face.
- Pricing model
- Not established
- Deployment
- Weights fit on a single 80 GB GPU in bfloat16; can be served with vLLM.
- Privacy
- Researchers can keep sensitive or unpublished sequences on their own infrastructure.
- Licence
- Apache-2.0
Capability
Modality
Access
Deployment
Pricing
Target user
Evidence
6 claims- Carbon-X is a 14.8B-parameter decoder-only language model trained jointly on DNA sequences and natural-language text. [1]
**Carbon-X** is a 14.8B-parameter decoder-only language model trained jointly on DNA sequences and natural-language text.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - It predicts whether mutations are benign or pathogenic, identifies regions that control gene activity, classifies sequences by organism group, and predicts next bases. [1]
You give it raw DNA inside an ordinary chat conversation and it answers in English: it predicts whether mutations are benign or pathogenic, identifies regions that control gene activity, classifies sequences by organism group, predicts the next bases of a sequence, and explains its reasoning.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - The model is available at HuggingFaceBio/Carbon-X with instructions for loading it using Transformers. [1]
- The weights fit on a single 80 GB GPU in bfloat16 and can be served with vLLM. [1]
The weights fit on a single 80 GB GPU in bfloat16, so researchers can keep sensitive or unpublished sequences on their own infrastructure.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Carbon-X averages 66.7% across five reported DNA benchmarks with thinking disabled. [1]
Across five DNA benchmarks, which cover variant-effect prediction, sequence understanding and next-base prediction, Carbon-X averages **66.7%** with thinking disabled.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights. - Researchers can keep sensitive or unpublished sequences on their own infrastructure. [1]
The weights fit on a single 80 GB GPU in bfloat16, so researchers can keep sensitive or unpublished sequences on their own infrastructure.
Hugging Face. Quoted text is outside CC BY 4.0 and keeps its original rights.
Sources
1
1
Hugging FaceOwner-controlled
https://huggingface.co/HuggingFaceBio/Carbon-X
Retrieved 2026-10-29 00:21 UTC
Verified
Versions
2| Version | Effective | Change | Changed fields | State |
|---|---|---|---|---|
| 2 | 2026-10-29 00:21 UTC | material change | capability, modality, pricing, availability | Current |
| 1 | 2026-10-27 00:11 UTC | publication | None | Current |
In briefs
1- Create expressive speech from a short voice sample Daily brief 2026-10-27
Original catalog content is licensed under CC BY 4.0, attributed to wowsuchai. Source quotations and linked content keep their original rights.