Handling Model Not Found, Rate Limiting, and Service Busy

404 Model Not Found (model not found)

Cause	What to do
Wrong model name	Go by the group model list on this site; don't copy from other platforms
The key's group doesn't include that model	Switch groups or switch models
Base URL is missing /v1	Add it, e.g. https://kuaiapi.net/v1
The model has been retired from the channel	Check the announcements and switch to an alternative model

429 Too Many Requests (rate limited)

A 429 isn't necessarily rate limiting. Check in this order:

1. Insufficient balance / quota → start with Insufficient balance
2. Request rate over the limit → lower your concurrency/rate and retry later
3. Account-level concurrency maxed out → wait for the requests already in flight to finish
4. Channel quota exhausted → retry at another time or contact support (support@kuaiapi.net)

TIP

Don't treat every 429 as rate limiting: insufficient balance, used-up plan benefits, and exhausted model quota all return 429.

5xx Service Busy (service busy)

• Fluctuations in the channel's model service, network jitter, or high load
• What to do: retry with exponential backoff (1s, 2s, 4s…); don't hammer it when you get an error
• If it happens on only one endpoint → switch to a recommended endpoint
• If it keeps failing → give support the time, model, and request ID (support@kuaiapi.net)

Three things to reduce errors

1. Lower your concurrency: cap the concurrency for long-running or batch tasks
2. Use the right model: use model names exactly as they appear in the group list
3. Read the announcements: the site admin usually posts before channel maintenance, price changes, or model retirements