Comet ML
ML experiment tracking and LLM observability platform (Opik), available open source, limited free plan, or paid subscription
Overview
Comet ML offers two complementary platforms: Opik, an open source observability and testing tool for LLM and agent applications (tracing, test suites, playground), and a classical MLOps platform for experiment tracking and model management. Opik self-hosts for free, or is used in the cloud with a Free plan (25,000 spans per month, 60-day retention, 10 members) and a Pro plan at $19/month (100,000 spans per month, 50 members, additional spans billed at $5 per 100,000 spans). The MLOps platform has its own pricing: Free at $0/month (100 GB of data, 1 user), Pro at $19/user/month (1,500 training hours, 500 GB storage, up to 10 users), and Enterprise by quote for unlimited users.
Comet ML is not translated into French: the website and documentation are in English only. The Enterprise plan documents SOC 2, ISO 27001, ISO 9001, HIPAA, and GDPR compliance, but these guarantees are not mentioned for Free and Pro plans. A complete API and SDKs cover most use cases, with official integrations to PyTorch, TensorFlow, Hugging Face Transformers, scikit-learn, XGBoost, and PyTorch Lightning. The main limitation is the dual pricing structure (Opik for LLMs, MLOps for classical machine learning) which can be confusing: a team with both use cases may need to purchase two separate Pro subscriptions.
Our verdict
Best for teams wanting to both track classical model training and monitor LLM applications in production under one brand. Not for you if you are looking for a single, simple pricing structure: Opik and Comet's MLOps platform are billed separately, which can double the subscription for a team using both.