Usage & plans

How metering, quotas, and plans work for your organization.

What is metered

Each calendar month (UTC), the platform records:

• Data read — rows and query counts from Explore, SDK data.query, and data_export jobs
• Compute — job runtime (factor_batch, data_export, strategy_validation)
• Storage — hosted Job artifact bytes retained for your org (platform MinIO)
• Research store — current bytes under your managed research prefix (tenants/{org}/ on the platform research bucket). Shown on Usage; not mixed with Job artifacts; no hard cap and no Stripe overage yet

Quotas

Free organizations have hard monthly caps on data-read rows, queries, compute, and storage. When a cap is hit, new queries or jobs are rejected (429 quota_exceeded) until the next month or until you upgrade.

Abuse protection is separate from monthly quotas. A burst may return 429 rate_limited with a Retry-After response header; wait for that interval before retrying. Upgrading a plan does not bypass this short-window protection.

For each metric that has a hard cap, the platform emails the organization Owner once when usage reaches 80% of that cap, and once again at 100%. The same threshold is not emailed again in that UTC month. Pro data-read rows and query counts have no hard cap, so they do not send these emails; Pro compute and storage still do. The email is a notice. It does not change the 429 that applies when a hard cap is already exhausted.

Pro still records data-read rows and query counts (visible on Dashboard and Usage) but does not hard-cap them — those meters will not block work. Pro keeps a hard monthly compute cap (100 hours by default) plus higher storage and concurrent-job caps.

data_export files on Artifacts are a short-lived platform cache (about 48 hours). If the cache grows past ~30 GiB, or the host disk is getting full, the oldest export files are deleted first. Download them to your research machine (or your own S3) if you need to keep them. Hosted factor_batch / strategy outputs are not part of this cache.

Your organization plan may come from Plans → Upgrade (Stripe) or from a platform operator (Admin → Customers). After a plan change, sign in again so the console session picks up the new tier.

Plans page

Open Plans to compare tiers and, as an Owner, start checkout or open the customer portal to update payment methods or cancel. Self-serve Upgrade requires production Stripe keys; contract customers are typically set to Pro by an operator.

Tips

• Prefer Explore with a limit for interactive single-day checks.
• Jobs data_export without a limit pulls a full trade date and, on Free, can consume a large share of the data-read quota. Export parquet on the platform is temporary — copy it off Artifacts if you need it.
• Production data_export reads live compacted data. Non-demo factor_batch fails.
• Delete unused artifacts if the Job-artifact quota is tight; expired export cache is also reclaimed automatically. Research-store objects are not this cache — they stay until you delete them.
• Coordinate seat / role changes in Settings so Member quotas stay within the org pool.