ai_ml_operations/chat_backend
2
0
Fork 0
Code Issues 23 Pull Requests Actions Packages Projects Releases Wiki Activity

Token-based billing, quotas, and usage enforcement #17

New Issue
Open
opened 2026-07-26 06:00:39 -07:00 by westfarn · 2 comments
westfarn commented 2026-07-26 06:00:39 -07:00
Owner
Copy Link
Copy Source

Follow-on to #15 (backend token tracking + admin panel).

Summary

Build billing / quota / enforcement on top of the token usage now tracked per turn. #15 only records tokens; this ticket acts on them.

Depends on

  • #15 — PromptMetric.tokens_in / tokens_out persisted with conversation-level aggregation.
  • #36 — multi-plan catalog (Founders / Standard / Pro / Business / Backer), feature flags, and prompt-window rate limits. Token-period quotas here should read limits from the same plan model #36 introduces (or extend it), not a second parallel plan table.

Related / coordinate with

  • #16 — chat UI token (and optionally plan/quota) display
  • #36 — owns plan definitions, Backer whitelist, feature gating (text vs image), and 100/200/300 prompts per 6h. This ticket owns token usage rollups over a billing period and warn/block based on those token caps.

Requirements

  • Define quota model (per user and/or per company; period-based, e.g. monthly) keyed off plans from #36.
  • Aggregate token usage per user/company over a billing period (reuse conversation-level rollups from #15).
  • Enforce limits: warn near cap, block or degrade when exceeded (define behavior).
  • Admin/reporting: view usage vs. quota; optionally reset/adjust quotas.
  • Consider provider-usage gaps: turns with null tokens must not silently under-count — document handling.
  • Document precedence vs #36 prompt-window limits (either gate may block a turn).

Acceptance criteria

  • Quota configuration per user and/or company (aligned with #36 plans)
  • Period usage aggregation from PromptMetric tokens
  • Enforcement path (warn + block/degrade) wired into chat flow
  • Admin visibility into usage vs. quota
  • Tests cover aggregation, threshold warnings, and enforcement
  • Shared plan→limits with #36; no duplicate catalog

Notes

Tracking-only is intentionally the scope of #15; billing/quotas/enforcement deferred here. Prompt-count rate limits and feature flags live in #36; this ticket is token-period billing/enforcement.

Follow-on to #15 (backend token tracking + admin panel). ## Summary Build billing / quota / enforcement on top of the token usage now tracked per turn. #15 only *records* tokens; this ticket *acts* on them. ## Depends on - #15 — `PromptMetric.tokens_in` / `tokens_out` persisted with conversation-level aggregation. - #36 — multi-plan catalog (Founders / Standard / Pro / Business / Backer), feature flags, and **prompt-window** rate limits. Token-period quotas here should read limits from the same plan model #36 introduces (or extend it), not a second parallel plan table. ## Related / coordinate with - #16 — chat UI token (and optionally plan/quota) display - #36 — owns plan definitions, Backer whitelist, feature gating (text vs image), and **100/200/300 prompts per 6h**. This ticket owns **token usage rollups over a billing period** and warn/block based on those token caps. ## Requirements - Define quota model (per user and/or per company; period-based, e.g. monthly) **keyed off plans from #36**. - Aggregate token usage per user/company over a billing period (reuse conversation-level rollups from #15). - Enforce limits: warn near cap, block or degrade when exceeded (define behavior). - Admin/reporting: view usage vs. quota; optionally reset/adjust quotas. - Consider provider-usage gaps: turns with null tokens must not silently under-count — document handling. - Document precedence vs #36 prompt-window limits (either gate may block a turn). ## Acceptance criteria - [ ] Quota configuration per user and/or company (aligned with #36 plans) - [ ] Period usage aggregation from PromptMetric tokens - [ ] Enforcement path (warn + block/degrade) wired into chat flow - [ ] Admin visibility into usage vs. quota - [ ] Tests cover aggregation, threshold warnings, and enforcement - [ ] Shared plan→limits with #36; no duplicate catalog ## Notes Tracking-only is intentionally the scope of #15; billing/quotas/enforcement deferred here. Prompt-count rate limits and feature flags live in #36; this ticket is token-period billing/enforcement.
westfarn commented 2026-07-31 04:01:43 -07:00
Author
Owner
Copy Link
Copy Source

Coordination note: Multi-plan / feature / prompt-quota work is #36.

Split of ownership:

  • #36 — plan catalog, Backer emails, feature gates (text vs image), prompts per 6h (100/200/300)
  • #17 (this) — token aggregation over billing period + warn/block on token caps

Depends on #36 for the shared plan→limits model. Either limit may block a generation turn; document precedence in the PR.

**Coordination note:** Multi-plan / feature / prompt-quota work is #36. **Split of ownership:** - #36 — plan catalog, Backer emails, feature gates (text vs image), **prompts per 6h** (100/200/300) - #17 (this) — **token** aggregation over billing period + warn/block on token caps Depends on #36 for the shared plan→limits model. Either limit may block a generation turn; document precedence in the PR.
westfarn commented 2026-07-31 04:22:38 -07:00
Author
Owner
Copy Link
Copy Source

Implemented in PR #37 (token-period quotas + aggregation) with FE usage display in ai_ml_operations/chat_web_app#72. Shares plan model with #36.

Implemented in PR https://git.aimloperations.com/ai_ml_operations/chat_backend/pulls/37 (token-period quotas + aggregation) with FE usage display in https://git.aimloperations.com/ai_ml_operations/chat_web_app/pulls/72. Shares plan model with #36.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
No items
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
twestfall westfarn (Ryan Westfall)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: ai_ml_operations/chat_backend#17