Union Benchmarks
Learn how Union scales with virtually no limit and enables you to saturate your compute.
Studies
Pick a study
Multi-cluster scale-out study
Orchestration without Limits
Union scales out to complete 200,000-action workflows where single-cluster OSS Flyte OOM-kills its executor.
200k actions, 0 failures
GPU-utilization study
Reuse or Reload
Container reuse keeps a 7B-model GPU at a sustained 100% utilization — 4.1× faster than spinning a fresh container per call.
4.1× faster
Coming soon
Have a workload you want benchmarked?
This is an evolving series — new studies land here as they ship.
Run these benchmarks yourself
Spin up a real Flyte cluster in minutes and run the same workloads on your own hardware. No infra to provision, no YAML to write first.