DagsHub
Free tierEverything you need to manage AI data & models in one platform
Free tier available·Technical·API available
Key strengths
Unified platform for dataset curation, experiment tracking, and model managementMultimodal data support (vision, audio, LLM) at petabyte scaleMLflow-compatible experiment trackingFull model lineage from model back to source dataOn-premise / VPC / air-gapped deployment options
Free tier + paid plans · from $99 USD/mo
Tel Aviv, Israel
Founded 2019
Self-hostable
No ratings yet
- MLflow-compatible experiment logging — integrate DagsHub as a remote MLflow tracking server to log runs, hyperparameters, metrics, and model artifacts from any ML framework (PyTorch, TensorFlow, scikit-learn, etc.).
- DVC-backed dataset versioning — version large datasets stored in cloud or on-premise storage with full lineage and reproducibility, using DVC-compatible pipelines.
- Multimodal auto-labeling pipelines — use the Team/Enterprise annotation workspace to run model-assisted or auto-labeling on vision, audio, and text data at scale.
- CI/CD/CT pipeline automation — connect repository events to trigger automated training, evaluation, and deployment workflows with interactive pipeline DAGs.
- Air-gapped enterprise MLOps — deploy DagsHub fully on-premise in a VPC or air-gapped environment with OpenShift support, SSO/LDAP/OIDC, and RBAC for regulated industries.
- Model registry with full lineage — register model versions in a central registry with traceability back to the dataset version, annotation run, and experiment that produced each model artifact.