Skip to main content
Your own provider keys don’t touch billing. Everything below applies to Managed Keys.

​
Managed Keys

Turn Managed Keys on for a provider and we run that provider’s requests on our own keys, so you don’t need an account with them. We require you to prepay a balance, which is then used to pay off model costs for the provider chosen. You turn Managed Keys on for each provider separately. OpenAI can run managed while Anthropic stays on your own key.
  • OpenAI serves gpt-* and o* models.
  • Anthropic serves claude-* models.
  • Gemini serves gemini-* models.
  • Workers AI serves the open-source families.
  • Bedrock serves claude-* models only.
  • Vertex serves gemini-* models only.

​
Refunds

Ask for a refund of unspent cash on the dashboard Billing page. It goes back to the card you paid with. We keep the top-up fee and can’t refund promo credits. Cash held by a request still in flight counts as spent until that request settles.

​
When the balance runs out

When the balance hits zero, managed requests stop and return 402 insufficient_balance until you add funds. Your own key routes keep working. We send a low-balance notice at $5, and auto-refill can top up for you. A daily spending limit also caps how much a runaway loop can spend in one day. A disputed card payment stops everything, your own key routes included. That’s 403 payment_disputed, and adding funds doesn’t lift it. Withdraw the dispute with your bank, or talk to us.

​
Cost inside a stream

A response you didn’t stream reports its cost twice, in the x-flexinference-cost header and the usage.cost block. A streamed response can’t use the header, so send include_cost: true to move the cost inside the stream.
The cost then goes in the usage frame the endpoint already sends. On /v1/chat/completions, also send "stream_options": {"include_usage": true}, because OpenAI makes that frame opt-in there. A managed request reports what the provider charged us. An own-keys request reports the provider’s list price at the tier that ran it. A request we can’t price reports no cost block at all.

​
Managing billing

You add a card, top up, set auto-refill, and ask for refunds on the dashboard Billing page.