What to Do When You Get "model not found" / "模型不存在" When Calling an OpenAI-Compatible API Proxy
Three Most Common Causes
- Model name doesn't match the list on this site. Proxy model names are not always identical to the vendor's official names. Client built-in dropdown menus often show the official names, causing mismatches when used directly.
- The model is not available in the group associated with the current token. 树新在线 uses default / vip groups to control model permissions, and different groups have different supported scopes.
- The client rewrites the model name or adds prefixes/suffixes. Some clients automatically prepend provider prefixes or alter the format, so the name actually sent differs from what the user entered.
How to Confirm the Actual Model Name
Go to https://treenew.online/models/ to view the full list. When copying a model name, it must match exactly, including case, hyphens, and spaces. Do not modify or translate it yourself.
How to Confirm What the Client Is Actually Sending
Enable client logging and check the model field in the request body; or use a packet capture tool to inspect the HTTP request; or run a manual curl test with the same request to compare results and verify the model name being sent.
Example curl command (replace YOUR_KEY and MODEL_NAME):
curl https://api.treenew.online/v1/chat/completions \
-H "Authorization: Bearer YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"MODEL_NAME","messages":[{"role":"user","content":"hi"}]}'Desired Model Not in Client Dropdown
Most clients support manually entering a custom model name. Type the exact name from https://treenew.online/models/ directly into the model field instead of relying on the dropdown list.
Still Getting Errors After Confirming the Name
If the model name matches the list exactly but the error persists, it is usually a group permission issue. Log in to the console at https://api.treenew.online, check which group your token belongs to (default or vip), and verify that the group supports the target model.