Evidently
Monitor ML and LLM systems: drift, quality, and evals.
About
Evidently is the open-source ML monitoring platform: detect data drift, track data quality, and evaluate both classical ML and LLM systems — with interactive reports and a monitoring UI. 8k stars, Apache-2.0, 823k PyPI downloads/month. It covers the 'is my model still behaving' layer that training tools ignore: reference-vs-production comparisons, test suites, and LLM-as-a-judge evals. The commercial Cloud gets the newest features first; the OSS core remains genuinely useful for drift and quality checks.
What it can do
- **Drift detection**: data and prediction drift out of the box
- **Data quality**: missing values, duplicates, schema checks
- **LLM evals**: quality, safety, and relevance metrics
- **Reports**: interactive HTML reports
- **Monitoring UI**: track metrics over time
Setup
```bash
pip install evidently
```
```python
from evidently import Report
from evidently.presets import DataDriftPreset
report = Report([DataDriftPreset()])
my_eval = report.run(reference_data, current_data)
```
Docs: docs.evidentlyai.com.
Prerequisites
Python 3.9+. Reference and current datasets to compare.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: 7,983 GitHub stars (log-scaled vs catalog peers); blended with ~823,148 monthly package downloads; last push 14d ago; 8 releases/12mo; 330 open issues. Manual review: ML/LLM monitoring (drift, data quality, evals), 8k stars, 823k PyPI/mo; Cloud gets newest features; crowded space. License: Apache-2.0.
★ 7,983 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.8/10
- Maintenance · recency, releases, issue load
- 7.2/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · 7,983 GitHub stars (log-scaled vs catalog peers)
- · blended with ~823,148 monthly package downloads
- · last push 14d ago; 8 releases/12mo; 330 open issues
- · manual review: ML/LLM monitoring (drift, data quality, evals), 8k stars, 823k PyPI/mo; Cloud gets newest features; crowded space
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.