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
Developer & Technical Documentation
DagsHub is designed to integrate with existing ML stacks with minimal friction:
- Experiment Tracking: Fully compatible with MLflow — point your
MLFLOW_TRACKING_URIto your DagsHub project to log runs, parameters, metrics, and artifacts without changing your training code. - Data Versioning: Built on open-source formats (DVC-compatible) for tracking dataset versions and lineage. Supports petabyte-scale data management at the Enterprise tier.
- Storage Integration: Connect your own AWS S3, GCS, or Azure Blob Storage. On the free tier, 20 GB of DagsHub-managed storage is included.
- Annotation Pipeline: Label Studio-compatible annotation workspace for multimodal datasets (vision, audio, LLM/text). Supports auto-labeling workflows on Team and above.
- CI/CD/CT Integration: Trigger continuous training pipelines from repository events. Interactive pipeline visualization is included on all tiers.
- Deployment: Enterprise tier supports deploying models to your own cluster, full VPC/air-gapped on-premise installation, OpenShift compatibility, SSO/LDAP/OIDC, and organizational RBAC.
- Access Control: Team-level RBAC available on Team tier; enterprise-grade SSO, LDAP, OIDC, and org-wide resource controls on Enterprise.