DagsHub
Everything you need to manage multimodal AI
Founded
2019
Starting Price
$0
Category
AI & Machine Learning
Last updated June 14, 2026
About DagsHub
DagsHub is a Git-based platform for data science and machine learning teams to version data, models, and experiments alongside their code in one place. Built on open standards like Git, DVC, and MLflow, it adds data versioning, experiment tracking, multimodal data annotation, and model management so teams can collaborate across the entire ML lifecycle.
Pros & Cons
Pros
- Consolidates code, data, models, and experiments in a single Git-based interface
- Built on open standards (Git, DVC, MLflow) so you avoid vendor lock-in
- Generous free tier for individuals and public/open-source projects
- Strong multimodal data support including image, audio, video, text, and 3D
- Label Studio-compatible annotation with AI-assisted labeling built in
Key Features
Data Versioning & Lineage
Track changes to large datasets, code, and models built on Git and DVC, with full auditable history and the ability to roll back to any previous version.
Experiment Tracking
A fully MLflow-compatible tracking server logs parameters, metrics, and artifacts, with interactive visualizations and comparison tools to evaluate model performance.
Multimodal Annotation & Auto-Labeling
Label Studio-compatible annotation workspace with AI-assisted labeling for image, audio, video, text, documents, tabular, and 3D/point-cloud data.
Data Engine
Curate, query, filter, and visualize petabyte-scale multimodal datasets with metadata versioning and visual metadata editing.
Model Management
Version, publish, and reproduce models, linking each model back to its experiment, data, and code, with model webhooks and deployment to your infrastructure.
Notebook Versioning & Diffing
Version and visually diff Jupyter notebooks so changes in exploratory work are tracked like any other artifact.
Pricing
Individual
$0/user/month
- Unlimited public repositories with unlimited collaborators
- Unlimited private repositories for non-commercial use
- Up to 100 tracked experiments in private repositories
- Up to 2 collaborators in private projects
- 20GB of DagsHub Storage
Best For
Reproducible ML experiments
Log parameters, metrics, and artifacts with an MLflow-compatible server and tie each run back to the exact data and code version used.
Multimodal data curation and labeling
Annotate and auto-label image, audio, video, and document datasets at scale with built-in Label Studio compatibility.
Data and model versioning for teams
Version large datasets and models with DVC while keeping a full auditable lineage of changes across the team.
Open-source ML collaboration
Host public repositories with unlimited collaborators to share data, models, and experiments with the community.
