Appendix: service limits

Limits below are taken from the live Flask order paths and storage policy. Prefer this page over older public API markdown that still documents a 15-HUC12 cap and HUC8 as a paid-only path. Plans are Free, Starter, MAX and Department; "MAX/Department" below is the server’s effective_pro gate, which Starter does not pass.

Storage-first tiers

• Free and Starter default storage: 2 GB (SWATGENX_BASIC_STORAGE_LIMIT_BYTES).
• MAX and Department default storage: 20 GB (SWATGENX_PRO_STORAGE_LIMIT_BYTES).
• Free and Starter retention: 30 days after creation (SWATGENX_BASIC_RETENTION_DAYS); MAX and Department have no scheduled cleanup.
• Concurrency: Free and Starter 1 build at a time; MAX and Department up to 3 (extras queue). Env: SWATGENX_BASIC_MAX_CONCURRENT_BUILDS / SWATGENX_PRO_MAX_CONCURRENT_BUILDS.
• Over storage quota → new orders blocked for any basin size (reason storage_limit).

Explorer outlet-HUC12 size cap (Free and Starter)

• POST /api/model-settings/explorer-watershed: Free and Starter accounts reject resolved watersheds larger than SWATGENX_BASIC_MODEL_MAX_HUC12 (default 35) with error_code model_basic_huc12_cap.
• The same gate message notes the HUC8 order path as the way to take one full HUC8 on any plan.
• All tiers: watershed must lie in a single VPUID (error_code model_cross_vpuid); cross-VPU basins can be explored but not ordered yet.
• MAX/Department explorer safety ceiling for very large single-region walks defaults to 300 HUC12 (SWATGENX_EXPLORER_MAX_WATERSHED_PRO), clamped 16–500.

Station, HUC8, HUC14 paths

• POST /api/model-settings (USGS station): storage + fair-use quotas apply; the old 15 contributing-HUC12 hard block is not applied on this path (huc12_count is still recorded).
• POST /api/model-settings-huc8: open to every plan — storage quota is the limit.
• HUC14 / catchment orders: available on every plan; storage_gate is the constraint.

Fair use, and MAX/Department features

• Free: up to 5 SWAT+ model orders per UTC calendar day (BASIC_DAILY_MODEL_LIMIT) and a lifetime allocation of 3 completed models on an account without a subscription (BASIC_LIFETIME_MODEL_ALLOCATION); failed builds are refunded.
• Calibration / validation flags on model creation (run inside the build) are MAX/Department only (code calval_pro_only). On-demand calibration from Calibration & Validation runs on cloud compute on every plan, paid from calibration credit.
• SWAT+MODFLOW-6: MAX/Department + groundwater coverage (Michigan Lower Peninsula, Indiana, Minnesota, Ohio, Wisconsin) via validate_modflow_addon — see the MODFLOW coupling chapter. Advanced QSWAT knobs follow MAX/Department; the resolution choice is open to every paid plan (Starter, MAX, Department).