Usage and billing
Understand Codius five-hour, weekly, and monthly usage windows, organization pooling, wallet credit, and plan changes.
Codius Coding Plans provide hosted model usage to an organization. Usage is pooled across organization members and API keys and is enforced through three simultaneous windows.
Included usage windows
Every admitted request must fit the active:
- five-hour window;
- weekly window;
- monthly entitlement period.
The Usage page shows the percentage consumed and exact reset time for each
window. The API returns the same view from GET /v1/codius/usage and through
usage headers on inference responses.
Codius intentionally presents comparable percentages instead of one universal token allowance. Different models and requests have different input, cached input, output, reasoning, context, and capability costs.
Pro and Max
Max receives exactly five times the corresponding Pro included-usage limits. The plans also differ in concurrency, organization-member, and active-API-key limits.
| Limit | Pro | Max |
|---|---|---|
| Concurrent requests | 3 | 10 |
| Organization members | 5 | 25 |
| Active API keys | 10 | 50 |
| Included usage | Base | Exactly 5× Pro |
Unused included usage never rolls over. Annual plans are billed once per year but still receive independent monthly entitlement periods.
Extra-usage wallet
Extra usage is opt-in. Included usage is consumed first; only the uncovered portion uses organization wallet credit.
- Wallet credit persists across monthly resets.
- Spending requires an active paid subscription.
- Top-ups are available in
$10,$25,$50, and$100amounts. - Auto-reload can be configured with a threshold, reload amount, and monthly cap.
- Reservations temporarily reduce the displayed available balance until a request settles.
Plan changes and cancellation
Upgrades and monthly-to-annual changes use a payment preview and take effect after the signed billing event is processed. Downgrades and annual-to-monthly changes ordinarily take effect at the paid period boundary. Cancellation is at period end.
Paddle remains authoritative for payments. Codius grants service access from signed webhook state, not from a browser redirect.
Quota responses
When a request cannot be admitted, the API returns an OpenAI-style 429 with a
stable Codius error code and reset or retry information. Do not retry
aggressively; wait for the supplied window or concurrency lease.
Last reviewed 2026-07-28 · Capability verified · Public docs are adapted and capability-checked before publication.
Provenance: apps/web/content/docs/models/usage-and-billing.mdx from CodiusAI/codius at working-tree.