> ## Documentation Index
> Fetch the complete documentation index at: https://docs.jiekou.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Rate Limits and Performance

## 1. Do image/video generation models have RPM limits?

Yes. RPM (requests per minute) limits vary by model:

* Veo 3.1: approximately 50 RPM
* Seedance: approximately 200 RPM
* Kling: depends on the specific model

## 2. Can I request an RPM limit increase?

Yes. RPM can be flexibly upgraded based on your usage needs. Please contact us and let us know your requirements.

## 3. What happens if actual usage does not reach the committed RPM?

If a user's actual RPM remains below the committed level for one consecutive week, the platform will adjust the rate limit according to the following policy (whichever is lower):

* Reduce the limit to the peak RPM over the past week
* Restore it to the model's default rate limit

## 4. What should I do if 429 "server overload" or "RPM limit exceeded" is returned?

This means the rate limit has been triggered. For model-specific limits, please refer to: [https://docs.jiekou.ai/docs/model/llm-rate-limits](https://docs.jiekou.ai/docs/model/llm-rate-limits)

Temporary solutions:

* Reduce the request frequency
* Add exponential backoff retries in your code
* Contact customer support to increase your RPM quota based on your recharge amount

## 5. Model responses often time out (Timeout)?

This is especially common with image editing models such as Gemini, which have longer processing times (possibly 1–10 minutes). Recommendations:

* Set the timeout to more than 10 minutes
* Use asynchronous task mode (if applicable)
* Avoid calling heavy tasks during peak hours

***

**Contact Support**

If the FAQs above do not resolve your issue, please contact technical support through the following methods:

* Enterprise WeChat/WeChat technical support group (recommended, fastest response)
* Information format to provide:
  * Issue description + screenshots
  * Account ID (UUID)
  * Trace ID (if any, usually in the error message)
  * Request parameters (after masking sensitive information)