Aim
Open-source, self-hosted experiment tracker with a fast UI for comparing runs.
About
Aim is an open-source, self-hosted AI metadata tracking tool designed to handle hundreds of thousands of tracked metric sequences. Developers log hyperparameters and metrics with two calls (aim.Run() and run.track()), then explore and compare runs in a performant web UI launched with aim up. Its pythonic query language supports searching, grouping, and aggregating experiments.
The project reports over 10 million downloads (vendor-reported) and is maintained by AimHub, which launched AimHub for Teams in May 2025, adding multi-project and organization support on customer premises with full compatibility with the open-source tracker.
Caveats: Aim was originally designed around single-user, single-project workflows, so team features are newer than the core tracker. Its ecosystem of integrations is smaller than Weights and Biases or MLflow, and very large-scale deployments need care around the self-hosted backend.
What it can do
- Metric and hyperparameter tracking
- Fast run-comparison UI
- Pythonic experiment search
- Group and aggregate metrics
- Self-hosted with local data ownership
Setup
pip install aim, add aim.Run() and run.track() calls to training code, then run aim up for the UI.
Prerequisites
Python 3.8+; training code to instrument.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 82; Trending score 72/100 from researcher assessment.
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.