Comet

Comet is a web platform for AI developers, researchers, and enterprise teams that supports end-to-end model evaluation across LLM applications and traditional ML workflows.

Key capabilities include:

• LLM evaluation and observability (Opik), including trace logging to capture and debug LLM calls, automated eval metrics (for example hallucination, context precision, relevance), and online evals for monitoring production data.
• Experiment tracking (MLOps platform) to log training runs, compare results, and support reproducibility across common ML stacks.
• Production monitoring to detect issues such as drift and investigate model behavior in deployed environments.
• Optimization and automation, including automatic prompt optimization runs for agentic systems and HyperTune.

Pricing: Freemium, Free plan ($0), Pro plan ($39/month; $39/user/month for MLOps), Enterprise (custom pricing).

Notable for combining open-source Opik LLM tracing and evaluations with an MLOps platform that covers experimentation and production monitoring, with options to use cloud or self-hosting for Opik.