Storage Cost

Published workflow outputs are stored in S3, which costs money for as long as they exist. Workflow Tracker estimates this cost wherever it shows an output size — the Dashboard summary cards, the runs list, and the Cost & Performance tab on a run.

Why it's shown as a range

We store outputs in buckets with S3 Intelligent-Tiering enabled. Objects automatically transition to cheaper storage tiers based on how recently they've been accessed — you don't pay for the more expensive tier on data you aren't reading. Because the actual cost depends on each object's access pattern over the year, Tracker shows a range: the low end assumes everything has aged into the cheapest tier, the high end assumes everything stays in the most expensive tier.

Tiers

Tier	When objects land here	Price (per GB/mo)	Annual (per GB/yr)
Frequent Access	First 30 days, or any object read recently	$0.021	$0.252
Infrequent Access	After 30 days with no access	$0.0125	$0.150
Archive Instant Access	After 120 days with no access	$0.004	$0.048

The dashboard range spans Archive Instant (low) to Frequent Access (high). A run whose outputs are never read again drifts toward the low end over its first few months; a frequently-read run stays near the high end.

Units

AWS labels storage sizes and prices in "GB/TB" but bills on binary (1024-based) units — what it calls a GB is really a GiB. Tracker follows the same convention, so the sizes shown alongside these costs are base-2.

Reducing storage cost

The biggest lever is deleting outputs you no longer need. The Dashboard supports bulk-selecting runs and deleting their outputs, and the CLI offers tracker outputs delete for individual runs. Deleting outputs zeroes the stored size and removes the ongoing storage charge.