mlop
mlop.ai
- Free plan
- Web
- Self-hosted
What mlop is
mlop is an open-source experiment tracking platform for machine-learning teams, available on the web or self-hosted. It records and compares experiments, model versions, hyperparameters, metrics, artifacts and gradients. Real-time monitoring, interactive dashboards and comparison views help teams follow training. Alerts can flag poor performance and suggest responses, including ways to address exploding gradients. The platform tracks experiment versions and uncommitted files as they change, supporting reproducibility. mlop states full compatibility with the Weights & Biases API and provides migration guidance for PyTorch Lightning and Hugging Face Transformers. A Python SDK is installed with pip. The Free plan costs nothing for one seat, unlimited logging hours and 10 GB storage.
Who it suits
It suits individual developers and small teams training machine-learning models, as well as organizations needing experiment comparison and reproducibility. Self-hosting is available on Enterprise.
Good to know, and what to watch
- Tracks experiments and model versions
- Real-time training monitoring
- Weights & Biases API compatibility
- Free plan includes unlimited logging hours
- Self-hosted Enterprise option
- Compute is labeled Private Beta
- Inference is listed as Coming Soon
- Free plan provides one seat
Plans
Every plan and priceEnterprise
Pro
Free
Free plan
The verdict
mlop focuses on experiment records, training visibility and reproducibility with web and self-hosted deployment. Its free tier supports individual work, while larger teams must discuss paid plans.
What the maker publishes
- API key security
- The documentation says secure API keys are hashed and shown only when generated, while insecure keys are not hashed. source
- Availability
- The homepage labels Compute as Private Beta and Inference as Coming Soon. source
- Contact
- The contact page offers a 15-minute consultation by Google Meet or Zoom. source
- Experiment tracking
- The platform tracks and compares experiments, model versions, hyperparameters, metrics, artifacts, and gradients. source
- Framework integrations
- The documentation provides migration instructions for PyTorch Lightning and Hugging Face Transformers. source
- Insights and alerts
- mlop says it can alert users to poor training performance and provide suggestions, including how to address exploding gradients. source
- Intended users
- The documentation describes mlop as a tool for people who train machine learning models and says its free tier is for individual developers and small teams. source
- Product
- mlop describes itself as an open source experiment tracking platform for machine learning teams. source
- Reproducibility
- The maker says the platform tracks experiment versions and uncommitted files in real time. source
- SDK
- The quickstart instructs users to install the Python SDK with pip install "mlop[full]". source
- Security option
- The Enterprise plan lists a security audit and a self-hosted option. source
- Support
- The pricing page includes email support for Pro and 24/7 support from founders for Enterprise. source
- Visualization
- Its documentation describes real-time training monitoring, interactive performance dashboards, and experiment comparison views. source
- Weights & Biases compatibility
- The site says mlop is 100% compatible with the Weights & Biases API, and the documentation describes migration support. source
Checked against the maker's pages on 8 Oct 2026.