Skip to main content
HaiToken API employs a dynamic rate limiting strategy that adjusts in real-time based on your account balance, system load, and channel quotas, ensuring a stable and fair service experience for all users.

​
Concurrency Limits

Actual available concurrency is dynamically adjusted based on the following factors:
Concurrency refers to the number of requests currently being processed at the same time. When existing requests have not yet returned results, new requests will consume concurrency quota.

​
TPM and RPM Limits

Currently, there are no fixed hard limits on TPM (Tokens Per Minute) and RPM (Requests Per Minute). Instead, a dynamic adjustment strategy is used:
  • System Load Awareness: Traffic is dynamically regulated based on the current overall system load
  • Channel Quota Coordination: Different model channels have different upstream quotas; requests are automatically throttled when quotas are exceeded
The dynamic adjustment strategy flexibly handles traffic spikes without affecting normal usage, while preventing any single user from exhausting system resources.

​
Frequently Asked Questions

Concurrency is dynamically adjusted based on your account balance. If your balance is low, the system automatically reduces the concurrency limit to control risk. Concurrency will be restored automatically after you top up.
Currently, rate limits are dynamically managed by the system based on balance and load, and manual adjustment is not supported. Maintaining a sufficient account balance ensures a more stable concurrency experience. For business needs requiring high concurrency, please contact customer service.

​
Next Steps