Follow these steps when you encounter issues — most common errors can be quickly identified.
⚡ Quick Self-Check List
Before contacting support, run through this checklist — 90% of issues can be found here:
Most Common Causes:
- Key and Base URL must be paired correctly: You cannot use a platform token to call the official OpenAI endpoint, and you cannot use an official OpenAI key to call our endpoint (ailink.nativex.com)
- Copy the Key in full: Extra spaces at the beginning or end will cause a 401 error
- Verify the model name character by character: A typo or unsupported model name will return 404 / no available channel
- Token quota and account balance are independent: Even if your account balance is sufficient, a token that has reached its usage limit will still report insufficient quota — check both "account balance" and "token remaining quota" separately
- Token status: Confirm in the console "Tokens" page that the token is not disabled or expired
Quota calculation formula:
Consumed quota = group multiplier × model multiplier × (prompt tokens + completion tokens × completion multiplier)
GPT-3.5 completion multiplier is fixed at 1.33; GPT-4 is fixed at 2.
🔍 Troubleshooting Steps
Work through these in order — the most common issues are listed first:
1. Check the error code and handle accordingly
Find the corresponding error code in the "Status Code Reference" table and follow the suggested action.
2. Try the example code to verify basic connectivity
Use a simple curl command to test, ruling out issues in your own code:
curl https://ailink.nativex.com/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer sk-your-token-key" \
-d '{
"model": "gpt-4o-mini",
"messages": [{"role": "user", "content": "hi"}]
}'
If the example works, the issue is in your business code. If it also fails, continue to the next step.
3. Switch models to check if the issue is model-specific
Replace the current model with another (e.g. gpt-4o-mini) and retry. If a different model works, the original model may be temporarily unavailable or your token may not have permission for it.
4. Create a new token Key to rule out token issues
Create a temporary token in the console and retry with the new Key. If the new Key works, the original token may be expired, disabled, or out of quota.