FAQ and troubleshooting

Find the cause from the error: 401, 403, 404, model not found, out of credit, timeouts.

Try the connection check first: pick your tool, paste the address you entered and find the failing step in one click. Or make sure the key and address work on their own, then look at the tool's settings. If the command below returns a model list, the key is valid and the site is reachable, and the problem is in the tool configuration.

cURL

curl https://tokenfunc.com/v1/models -H "Authorization: Bearer sk-xxxxxxxxxxxxxxxx"

PowerShell

Invoke-RestMethod -Uri "https://tokenfunc.com/v1/models" -Headers @{ Authorization = "Bearer sk-xxxxxxxxxxxxxxxx" }

Find the cause from the error

Error	Usual cause	What to do
`401` invalid token	Key copied incompletely or with spaces; key deleted, expired or out of its own quota	Check the key's status and remaining quota on API Keys and copy it again
`403` token has no access to the model	The key limits its models, or its group does not include the model	Allow the model in the key settings, or use a key from another group
`403` insufficient user quota	Account balance is too low	Top up under Wallet; requests work again right away
`404` Not Found	One `/v1` too many or too few in the base URL	Check the address against the tool guide, see Quick start
`503` no available channel / `model_not_found`	Wrong model name, or the model is temporarily unavailable	Copy the name from the Model Square; try another model
`429`	Too many requests	Lower concurrency and retry later
`413`	Request body too large (usually a big image or file)	Compress the image or send an image URL instead
Timeout or no response	Very long output or an unstable network	Turn on streaming; check your proxy settings

More status codes and retry advice are on Errors.

FAQ

Does the base URL include /v1 or not?

It depends on the protocol the tool uses: OpenAI-compatible tools want /v1, Anthropic-protocol tools such as Claude Code do not. Every tool guide says which; when in doubt, see the table on Tool integration overview.

The model is on the Model Square, so why does the call say I have no access?

The Model Square lists every model on the site. Which models a key can use depends on its group and on any model limits set on the key. List models returns what your key can actually use.

How am I billed, and where do I see usage and charges?

Charges come out of your balance by actual usage: most text models bill per token with separate input and output prices, and some bill per request. Each model's price is on the Model Square. Cached input is billed at a lower cache price; see prompt caching in Anthropic Messages.

Usage Logs in the console lists every request with its model, token counts and cost. Requests are rejected when the balance runs out and work again right after a top-up.