Carbon
Hugging Face's family of open generative DNA language models (500M, 3B, 8B) with 6-mer tokens and up to ~786 kbp context; zero-shot variant-effect scoring and sequence generation.
| Category | Genomics & DNA models |
|---|---|
| GitHub | huggingface/carbon |
| First public | 2026-05-19 (source) |
| Paper | https://github.com/huggingface/carbon/blob/main/tech-report.pdf · Technical report (2026) |
| License | Apache-2.0 |
| Pricing | Open source |
| Tags | #dna-model#foundation-model#generative |
| Last checked | 2026-10-17 |
Specs
| Input length / context | Up to ~393 kbp (3B) / ~786 kbp (8B) |
|---|---|
| Resolution | — |
| Species | — |
| Tasks | DNA generation, zero-shot variant-effect prediction |
| GitHub stars | 222 |
| Interface | — |
Specs come from each project's README, paper or official page · Carbon site · Carbon README. “—” means we couldn't verify it.
Compare with…
Pick 1–3 other Genomics & DNA models tools to compare side by side with Carbon.
Related
Predicts gene expression from sequence with long-range interactions.
Genomics & DNA modelsOpen source#expression#regulatory
Genome language model across all domains of life with 1M-token context.
Genomics & DNA modelsOpen source#dna-lm#foundation-model
DeepMind model predicting regulatory effects of DNA variants at base resolution.
Genomics & DNA modelsFree (research)#regulatory#variant-effect