Wafer 402 insufficient_credits: "Your credit balance is too low for the request"

Disclosure: ernd partners with Wafer.

What does the error mean?

Wafer’s error table lists 402 with code insufficient_credits and the meaning Your credit balance is too low for the request. Its retry column says to retry after a top-up.

Serverless on Wafer is prepaid. Before a request runs, Wafer works out its maximum possible cost and compares it with your balance. The estimate assumes the request generates its full max_tokens, multiplied by n. The error body carries credits_available_cents and credits_required_cents_estimate, so you can see the gap.

Wafer says the wording of message can change, so match on the code field.

How do you fix it?

1. Add credits at app.wafer.ai.
2. Retry the request. Requests rejected before generation starts aren’t charged, so the failed one cost nothing.
3. If you can’t top up right now, set a smaller max_tokens. Without one, Wafer uses the model’s default output length for the estimate, which can be large.
4. If the code is member_spend_limit instead, the account has a spend limit for that member, and whoever manages the account’s limits has to raise it.

Don’t loop on the 402. Wafer’s docs say to add credits first.

How do you keep it from coming back?

Wafer’s docs don’t describe auto-reload or low-balance alerts, so watch the balance yourself. The dashboard at app.wafer.ai shows it, and GET https://api.wafer.ai/v1/usage returns tokens and cost per model that your monitoring can poll.

Always send max_tokens. It keeps the up-front estimate close to what you really use, so a modest balance covers more requests.

Pay for credits on a card you keep for AI vendors. ernd cardholders earn 10% back on Wafer spend.