PyOD is a free Python library for anomaly detection, with 61 documented detectors across tabular, time-series, graph, text, image, and audio data. A shared API supports the classic detector workflow, while ADEngine adds orchestration: it profiles data, chooses benchmark-backed detectors, runs them in parallel, calculates consensus scores, and reports diagnostics. ADEngine also works as a Python API without an LLM. PyOD additionally provides an agentic investigation workflow and activation paths for Claude Code, Codex, and MCP-compatible agents. Optional extras enable capabilities such as PyTorch detectors, graph models, embeddings, audio, and an MCP server. Install through pip, conda-forge, or from source; the documented requirement is Python 3.9 or higher. PyOD serves academic research and commercial products. Its documentation cautions that ADEngine’s quality verdict is heuristic rather than a guarantee, and recommends checking results with held-out labels or domain review. The model persistence guide also warns that pickle and joblib can run arbitrary Python code when loading artifacts, so callers must pass trusted=True.
Who it is for
PyOD suits Python users working on anomaly detection across the documented data types, including academic and commercial projects. ADEngine may suit teams seeking orchestration, while optional agent integrations are available for compatible workflows.
What is good
- Documents 61 detectors across six data types.
- ADEngine runs detectors in parallel and computes consensus scores.
- Available through pip, conda-forge, or source.
- ADEngine works without an LLM.
- Free and open source.
What to know first
- Requires Python 3.9 or higher.
- Optional capabilities require pip extras.
- ADEngine’s quality verdict is a heuristic, not a guarantee.
- Loading pickle or joblib artifacts requires trusted=True.
Verdict
PyOD offers a broad detector catalog with both a classic API and an orchestration layer. Treat ADEngine’s quality verdict as a starting point and validate results; load serialized artifacts only when trusted.
PyOD plans and pricing
All plansCompared on anomaly detection software
- Free plan
- Yes
- Detection method
- hybrid
- Real-time detection
- Yes
- Supported data
- tabular, time series, graph, text, image, audio
- Deployment options
- self-hosted
- Anomaly explanations
- Yes
