Determined AI
Open-source deep learning training platform with distributed training and HPO.
Acquired — still operating
acquired by HPE Jun 2021
About
Determined is an open-source deep learning training platform that makes building models fast and easy. It bundles distributed training (no model-code changes required), state-of-the-art hyperparameter search from the creators of Hyperband, smart GPU scheduling that cuts cloud costs via preemptible instances, and experiment tracking covering code versions, metrics, checkpoints, and hyperparameters — exposed through a CLI, REST API, Python SDK, and web UI.
Determined AI was acquired by HPE in June 2021 and now sits in HPE's HPC and AI business unit, deployable on-premises on Kubernetes via HPE Ezmeral. The honest caveat is momentum: the project reads as a mature, stable platform rather than a fast-moving one, and teams evaluating it in 2026 should confirm the release cadence fits their needs. For shops that want a batteries-included training platform without assembling Ray + Tune + tracking themselves, it remains a solid, genuinely open option.
What it can do
- Distributed training without changing model code
- Hyperband-based hyperparameter search
- Smart GPU scheduling with preemptible instance support
- Built-in experiment tracking (code, metrics, checkpoints)
- CLI, REST API, Python SDK, and web UI
Setup
pip install determined, then det deploy local cluster-up for local runs or deploy on Kubernetes on-premises.
Prerequisites
Python, Docker (for local cluster), GPUs for training workloads.
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 77; Trending score 68/100 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.