Comet

Comet is an MLOps platform for experiment tracking, model registry, and production monitoring, letting ML teams log, compare, and reproduce runs with a free tier and per-seat paid plans.

Comet is a long-established MLOps platform that gives machine learning teams a system of record for their model development. Its core is experiment tracking: engineers instrument their training code to automatically log metrics, hyperparameters, source code, datasets, and artifacts, then use Comet's UI to visualize, compare, and reproduce experiments. This addresses one of the most persistent pain points in ML, keeping track of what was tried and why one model outperformed another. Beyond tracking, Comet provides a model registry to version and stage models toward production, along with production monitoring in its higher tiers to watch for drift and performance issues once models are deployed. It is framework-agnostic and integrates with common ML libraries, fitting into existing training pipelines with minimal changes. Comet offers a free MLOps tier for individuals and small projects with generous storage and fair-usage limits, a Pro plan around $19 per user per month with higher limits and up to ten users, and an enterprise tier with unlimited usage, advanced monitoring, flexible deployment (including self-managed options), SSO, and dedicated support. Comet also develops Opik, its separate open-source GenAI observability and evaluation product, but the core Comet platform remains focused on classic MLOps and experiment management.