Comet ML

What is Comet ML?

Comet is an MLOps and LLM observability company with two platforms: an MLOps platform for experiment tracking and model production monitoring, and Opik, an open-source platform for AI observability and agent testing.

Key capabilities of Comet ML

• Experiment tracking: logs and compares training runs
• Model registry: stores and versions models
• Production monitoring: available on Enterprise MLOps
• Opik agent tracing: traces and analyzes LLM agent behavior
• Test suites and assertions: tests LLM applications, even on free plans
• LLM evaluation: included across tiers

How Comet ML works

Teams add Comet to training code to log metrics and artifacts, then compare experiments and register models. For LLM apps, Opik captures traces and spans from agents and runs test suites with assertions. Opik can be self-hosted from the open-source release or used as cloud service, with span limits and retention set by plan.

Who uses Comet ML?

ML engineers, data scientists and teams building LLM agents use it. Academic users can apply for free Pro access on both platforms, and Enterprise suits organizations needing SSO and compliance.

Comet ML pricing

MLOps: Free for 1 user with 100GB storage, Pro $19 per user a month for up to 10 users with 1,500 training hours, Enterprise custom. Opik: open source $0, Free Cloud 25k spans a month, Pro Cloud $19 a month for 100k spans, Enterprise custom.

Comet ML alternatives

Alternatives include ClearML for open-source MLOps, BentoML for model serving, Seldon for Kubernetes deployment, Valohai for ML pipelines, and Domino Data Lab for enterprise model operations.