scGPT vs Geneformer
scGPT: Generative pretrained transformer foundation model for single-cell multi-omics. Geneformer: Transformer pretrained on ~30M single-cell transcriptomes for network biology predictions.
Facts below come from each project's own website, repository and paper. We don't publish benchmark claims here; see the papers for evaluations. Shared tags: foundation-model.
| Field | scGPT | Geneformer |
|---|---|---|
| Pretraining corpus | 33M human cells (whole-human checkpoint) | ~30M (original) / ~104M transcriptomes |
| Architecture | Generative transformer | Transformer |
| Tasks | Annotation, integration, reference mapping, perturbation | In silico perturbation, cell + gene classification (fine-tuning) |
| Species | Human | Human |
| License | MIT | Apache-2.0 |
| Pricing | Open source | Open source |
| Released | 2023-05-01 | 2023-05-31 |
| Paper | Link | Link |
| GitHub | GitHub | — |
| GitHub stars | 1,654 | — |
| Interface | Package | — |
Specs come from each project's README, paper or official page · scGPT README · Geneformer site · Geneformer paper. “—” means we couldn't verify it.
Add more Single-cell & foundation models tools to this comparison →