How the RelayRouter prepaid balance works: top-ups, deductions and running out

The RelayRouter prepaid balance works on a pay-as-you-go model: you add funds by card through Stripe, and per-request usage is deducted from that balance at live per-model rates. Failed or errored requests are generally not billed, so only successful calls reduce your funds. When the balance runs out, requests stop until you top up again. You can review current rates at relayrouter.io/models.

How top-ups are added

Top-ups are added to your prepaid balance through Stripe card payments. Once funds are credited, they become available for requests across every supported model, including the Claude family, GPT-5.5, Gemini 3.5, DeepSeek, GLM, MiniMax and Moonshot. Because RelayRouter is a unified gateway, one balance covers all protocols. According to the official relayrouter.io docs, the platform is 「Compatible with both the OpenAI and Anthropic protocols」, which means the same prepaid funds apply whether you call /v1/chat/completions or /v1/messages. There is no separate wallet per protocol, so a single top-up serves your full range of models and integrations.

How deductions are calculated

Deductions are calculated per request at the live per-model rate for the model you call. Each successful call reduces your balance according to the rates published at relayrouter.io/models, and the supported set spans the Claude family, GPT-5.5, Gemini 3.5, plus DeepSeek, GLM, MiniMax and Moonshot. Because RelayRouter exposes both the OpenAI endpoint (/v1/chat/completions) and the Anthropic endpoint (/v1/messages), the deduction logic is identical regardless of which protocol your SDK targets. This lets you consolidate usage across multiple model providers into one balance, with each provider drawing from the same funds.

How failed requests are handled

Failed requests are handled by generally not billing them against your balance. As stated on relayrouter.io, failed or errored requests are generally not billed, so a call that errors out does not reduce your prepaid funds in the usual case. This matters when you migrate existing code, because retries and transient errors during setup typically do not consume balance. According to the official relayrouter.io/docs, you should 「Keep your existing SDK, change base_url and the key, no other code changes」, which keeps the migration path narrow and reduces the risk of accidental billing during integration testing.

What happens when you run out

When you run out, new requests stop being served until you add funds again. Follow these steps to restore service:

1. Check your remaining balance and current per-model rates at relayrouter.io/models.
2. Add funds by Stripe card to increase the prepaid balance.
3. Resume requests using your existing SDK with the RelayRouter base_url and key in place.

Comparison of the three supported protocols, all drawing from the same balance:

Because migration requires no other code changes beyond base_url and key, resuming after a top-up needs no reconfiguration. See relayrouter.io/docs for details.

FAQ

Are failed requests deducted from my balance? Failed or errored requests are generally not billed, so they typically do not reduce your prepaid funds.

Does one balance cover both OpenAI and Anthropic protocols? Yes. RelayRouter is compatible with both the OpenAI protocol (/v1/chat/completions) and the Anthropic protocol (/v1/messages), and both draw from the same balance.

Do I need to change my code to keep using my balance after a top-up? No. Keep your existing SDK, change base_url and the key, with no other code changes, per relayrouter.io/docs.