Per-User Credit Usage Report — dev / stage / prod

Generated: 2026-08-24 Source: user_credit_ledger + user_credit_accounts collections, read directly from each environment’s MongoDB (via genesis-train’s database, database name test in all three — see credit-system-and-enforcement.md for why). Method: Read-only aggregation over the ledger’s deduction events (= actual usage) vs. grant/seed/refresh/purchase/manual_credit events (= credits added). Raw per-user CSVs for all three environments are in credit-usage-reports-2026-08-24/. Note on units: all figures below are in display credits (1 display credit = 1,000 millicredits, the internal ledger unit). See the companion doc for why the token→credit pricing behind these numbers is still placeholder-rate, not final pricing — that affects how these totals should be read (relative usage patterns are meaningful now; absolute dollar-equivalent cost is not yet, until real per-model rates are set).

Summary across environments

Environment	Users with ledger activity	Ledger transactions	Total credits spent (all users)	Total credits granted (all users)
dev	36	4,916	93,348.07	6,906,000.0
stage	62	5,785	94,680.12	6,246,000.0
prod	11	496	10,345.175	120,000.0

Read this carefully: dev and stage are dominated by internal team accounts (@inicreatechnologies.com) and synthetic test accounts (shanks101test+...@gmail.com, stripe.test@yopmail.com, etc.) with unrealistically large grants (some accounts show 1–2 million credits granted — clearly test/QA seeding, not real plan grants). Prod is the only environment that reflects real customer usage today, and it currently has just 11 users with any ledger activity at all.

Production — full detail (11 users)

Sorted by total credits spent, descending. Since enforcement is off in prod (see companion doc §4.3), “current balance” does not reflect any blocking behavior — these users have kept using the product freely regardless of balance.

User	Plan	Status	Credits spent	Deduction transactions	Credits granted	Current balance
john+3@altrucreative.com	Pro	Active	3,778.08	106	22,000.0	18,221.92
missfreq@icloud.com	—	—	2,815.575	125	10,000.0	7,184.425
stallingsaudio@gmail.com	Cohort Pro Plan	Active	2,216.15	103	26,000.0	23,783.85
budandthenight@gmail.com	Pro	Active	609.425	44	8,000.0	7,390.575
ericborlin@gmail.com	—	—	396.54	33	10,000.0	9,603.46
holtatkinson@gmail.com	Cohort Pro Plan	Active	154.19	16	8,000.0	7,845.81
goingaaronrogue@gmail.com	Cohort Pro Plan	Active	111.465	13	8,000.0	7,888.535
stallj89+070826@gmail.com	Pro	Active	104.555	13	8,000.0	7,895.445
castlemusic.ofc@gmail.com	Cohort Pro Plan	Active	77.785	10	4,000.0	3,922.215
geosmithtunes@gmail.com	Cohort Pro Plan	Active	72.89	8	8,000.0	7,927.11
hotboxxofficial@gmail.com	Promo Consult (Tier 1)	Active	8.52	1	8,000.0	7,991.48

Observations:

Heaviest prod user (john+3@altrucreative.com) has spent 3,778 credits over 106 AI actions — roughly 35.6 credits/action average, but this varies a lot by feature mix (chat vs. video/audio analysis cost very differently per the rate card).
Two accounts (missfreq@icloud.com, ericborlin@gmail.com) have no plan_name/subscription_status recorded but still show grants and usage — worth checking whether these are legacy or free-tier-adjacent accounts, since the companion doc notes there’s officially no free-tier grant path today.
No prod user is anywhere close to zero balance, so enforcement being off has not yet been “tested” by real usage patterns — nobody has hit the ceiling.

Full CSV: credit-usage-reports-2026-08-24/credit_usage_prod.csv

Dev — top 15 by spend (36 total, see CSV for full list)

User	Plan	Credits spent	Deductions	Credits granted	Balance
henrycikanek14@gmail.com	—	44,649.505	1,303	24,000.0	3,350.495
darshan.bajgain@inicreatechnologies.com	Plus Dev	24,371.13	734	2,010,000.0	1,985,628.87
service_user	—	7,015.42	319	10,000.0	2,984.58
locimol752@preparmy.com	Plus Dev	5,368.47	178	1,858,000.0	1,852,631.53
sanskar.karki@inicreatechnologies.com	Plus Dev	2,410.735	107	1,850,000.0	1,847,589.265
henrycikanek14+1@gmail.com	—	2,232.75	47	0	11,767.25
umesh@inicreatechnologies.com	Pro Dev Crazy	2,085.075	105	22,000.0	19,914.925
sanskarkarki20@gmail.com	Plus Dev	1,159.07	46	8,000.0	6,840.93
sharma.aatosh1416@gmail.com	Pro Dev Crazy	697.25	33	22,000.0	21,302.75
stallingsaudio@gmail.com	Pro Dev Crazy	408.62	23	22,000.0	21,591.38
bayawo9754@afterdo.com	Plus Dev	380.825	25	12,000.0	11,619.175
sapkotadilasha8@gmail.com	Plus Dev	324.81	16	4,000.0	3,675.19
rajib@inicreatechnologies.com	Pro Dev Crazy	309.43	27	116,000.0	115,690.57
jexifox638@cadebek.com	Plus Dev	196.28	24	8,000.0	7,803.72
pewow64183@bevriz.com	Plus Dev	188.305	22	8,000.0	7,811.695

Full CSV: credit-usage-reports-2026-08-24/credit_usage_dev.csv

Stage — top 15 by spend (62 total, see CSV for full list)

User	Plan	Credits spent	Deductions	Credits granted	Balance
bigyama.khanal@inicreatechnologies.com	Pro	26,416.53	1,191	30,000.0	3,583.47
shanks101test+stage4@gmail.com	Pro	13,379.995	368	22,000.0	8,620.005
shanks101test+stage2@gmail.com	Pro	11,231.945	398	20,000.0	8,768.055
sneha.panta@inicreatechnologies.com	Pro	6,165.49	254	2,022,000.0	2,015,834.51
shiraj@inicreatechnologies.com	Pro	5,121.26	171	1,346,000.0	1,340,878.74
shanks101test+stage@gmail.com	Pro	3,285.35	156	22,000.0	18,714.65
shanks101test+stage3@gmail.com	Pro	2,758.045	116	16,000.0	13,241.955
sapkotadilasha8@gmail.com	Pro	2,755.065	68	22,000.0	19,244.935
shanks101test+stage5@gmail.com	Pro	2,031.995	105	30,000.0	27,968.005
darshan.bajgain@inicreatechnologies.com	— (Canceled)	1,974.275	65	8,000.0	6,025.725
cikanekmgmt@gmail.com	Pro	1,590.45	65	22,000.0	20,409.55
koxab68956@gzeos.com	— (Canceled)	1,471.805	63	22,000.0	20,528.195
usopp101test@gmail.com	— (Canceled)	1,297.85	38	22,000.0	20,702.15
sanskar.karki@inicreatechnologies.com	Pro	1,265.675	46	12,000.0	10,734.325
bigyama97@gmail.com	Pro	1,235.8	70	1,336,000.0	1,334,764.2

Full CSV: credit-usage-reports-2026-08-24/credit_usage_stage.csv

Caveats for whoever uses this report

These numbers are only as meaningful as the pricing behind them. As documented in credit-system-and-enforcement.md §4.4, the token→credit rate table is explicitly placeholder (TODO: replace with researched rate), and two live models (gpt-audio, gpt-5.1) aren’t in the table at all and are billed at a generic fallback rate. Until that’s fixed, “credits spent” is a real relative signal (who uses the product more) but not yet a real cost signal (how much a user actually costs in USD).
This is a point-in-time snapshot (queried 2026-08-24). Usage is ongoing; re-run for a current view before making final limit decisions.
Dev/stage figures are not representative of real user behavior — they’re dominated by internal team members and synthetic QA accounts with artificially large grants (some in the millions of credits), so per-user averages from those environments should not be used to size real-user credit limits. Prod is the only environment to base limit decisions on, and it currently has a very small sample (11 users) — worth treating any limit decision as provisional until more real usage accumulates.
“Credits granted” includes test/QA seeding, not just legitimate plan grants — particularly visible in dev/stage where some users show 1–2 million credits granted via repeated refresh events, almost certainly from a test script re-running subscription-sync rather than real billing cycles.