Documentation Index
Fetch the complete documentation index at: https://docs.xhuoapi.ai/llms.txt
Use this file to discover all available pages before exploring further.
How do Credits convert to USD?
- The console shows prices in USD for reference
- Actual deductions are in Credits
- Each package has its own Credits ↔ USD ratio set at purchase time
When is billing applied?
| Call type | When billed |
|---|---|
| Synchronous API (Chat, Embedding, …) | At response time, by token / character / count |
| Async task (image / video / audio) | When the task reaches success |
Task final state failed | Generally not billed (per service detail page) |
| Gateway 5xx / upstream timeout | Not billed |
| 4xx auth / param errors | Not billed |
Content moderation rejection (forbidden / 403) | Not billed |
Instance balance vs general balance — which is used first?
- Instance balance (Dedicated) is tied to an application created under a specific service; usable only for that service
- General balance is shared across all services
- Calls draw from the instance balance first. If the instance application has “Allow Use General Balance” enabled, the call falls back to the general balance when the instance balance is exhausted.
What does the API return when balance is exhausted?
Are failed requests billed?
Not billed:- Gateway 4xx (
bad_request,invalid_token,used_up, …) - Upstream 5xx
- Gateway / upstream timeouts
- Content moderation rejections (
forbidden) - Async tasks ending in
failed
- Tasks finishing as
success - Successful synchronous responses
How do refunds work?
Pay-as-you-go packages generally do not support refunding consumed credit. Specific terms follow the package you purchased and any console announcements. For exceptional situations (e.g. extended service unavailability), email[email protected] with:
- Credential ID (do not send the full Token)
- Affected time window
trace_ids of the affected requests
How do I view usage?
The console Usage page lets you slice by service, time window, and credential. Each row carries atrace_id for pinpointing specific calls.
Can I share a Token across teammates / environments?
Not recommended. Reasons:- You can’t tell which consumer caused an issue
- A leak forces you to revoke the entire Token
- Usage is aggregated by Token — no per-consumer accounting
Related
Pricing overview
The pay-as-you-go model
Response format
All error codes and statuses

