Skip to content
Relaykit

FAQ · 21 answers

Questions, answered plainly.

Billing, models, agent setup, security and accounts. If your question is missing, email us and we will answer it and add it here.

Billing and credits

How does prepaid billing work?

You buy a credit pack, and each API request deducts its cost from your balance. One credit equals $1 of usage at the per-model rates on the models page. No subscription is required, and there is no seat fee or minimum monthly spend. Optional monthly plans (Pro $5, Plus $9, Max $29 a month) add a weekly allowance and are described on the pricing page. Plans are opening soon.

How is the cost of a request calculated?

Cost = input tokens × input rate + output tokens × output rate (+ cached tokens × cached rate where a model supports prompt caching), with rates quoted per million tokens. The usage log in your console shows the token counts and exact amount deducted for every request.

What is the smallest top-up?

The minimum deposit is $3, and models become available once an account has paid that much in total. Online checkout is not open yet: today you buy a credit pack by email from the billing page in the console, and packs start at $10. The price and credit amount are shown before you pay.

Do credits expire?

Purchased credits do not expire while your account is open and in good standing. Plan allowances refill weekly, and unused allowance does not carry over.

Can I get a refund?

Unused purchased credit can be refunded within the window described in the Refund Policy. Credit already spent on requests and promotional credit are not refundable. Email support@relaykit.dev with your transaction reference.

What happens when my balance runs out?

Requests return an insufficient-balance error (HTTP 403) until you top up. A request that has already started always finishes, so a long reply can take the balance slightly below $0; that amount comes off your next top-up.

Models and API

Which models can I use?

11 models from Claude, GPT, Grok and DeepSeek are available today. The models page lists every model ID, its rates and which API formats it supports.

Which API formats are supported?

Both. Use the OpenAI format at https://api.oneaiapikey.com/v1 (chat/completions, responses, models, embeddings) or the Anthropic Messages format by setting the base URL to https://api.oneaiapikey.com and calling /v1/messages. Anthropic-format requests work with claude-opus-5-5, claude-sonnet-5, claude-haiku-4-5 and claude-sonnet-5-5.

Why should I set stream: true?

Many networks and proxies close a connection that stays silent for about 100 seconds. A long non-streaming request can be cut before the model finishes, and you may still be charged for tokens already generated. Streaming sends tokens as they are produced and keeps the connection alive.

Which model answers my request?

The one you name. Each request is routed to the model ID you send, served by the provider that operates it, and that model ID is recorded next to the cost in your usage log.

Agent kits and tools

Which coding agents work with it?

Any tool that accepts a custom OpenAI-compatible or Anthropic-compatible base URL. The agent kits cover Claude Code, Codex CLI, Cursor, Cline, Kilo Code, OpenCode, Aider and Zed, each with a tested setup you can copy.

How do I use it with Claude Code?

Set ANTHROPIC_BASE_URL to https://api.oneaiapikey.com and ANTHROPIC_AUTH_TOKEN to your Relaykit key, then start Claude Code as usual. The Claude Code setup in the agent kits covers settings files and model selection.

Can I use non-Claude models in Claude Code?

Claude Code is designed around Anthropic models and the Anthropic format, so we recommend Claude models there. For GPT, Grok or DeepSeek, use a tool that speaks the OpenAI format, such as Codex CLI, Cline, Kilo Code or OpenCode.

Do the free tools need an account?

No. The cost calculator, agent config generator and token estimator run in your browser and work without an account or API key.

Security and privacy

Do you train on my prompts or code?

No. We never use your prompts or outputs to train models and we do not sell them. By default only request metadata (model, token counts, cost, time) is stored for your usage log. Requests are processed by the upstream provider under its own API terms.

How should I protect my API keys?

Treat keys like passwords: keep them in environment variables or a secrets manager, never in client-side code or public repositories. Create one key per app or teammate, set a spending cap on each, and revoke any key you suspect is exposed. Requests made with a leaked key are charged to your balance.

Is Relaykit affiliated with Anthropic, OpenAI, xAI and DeepSeek?

No. Relaykit is an independent API platform and is not affiliated with, endorsed by or sponsored by any model provider. Model names are trademarks of their owners.

Accounts and support

Is there a free trial?

No. There is no sign-up credit: new accounts start at $0, and models, including free models, become available once the account has paid $3 or more. The $3 trial gives the first month of Pro for one card payment, once per account. Plans are opening soon.

Can my team share one balance?

Yes. Top up one account and create a separate API key for each person or app, so you can see usage per key and revoke access individually. Sharing keys outside your team is not allowed.

How do I delete my account?

Email support@relaykit.dev from your account address. We delete or anonymise your personal data within 30 days, except records we must keep by law, as explained in the Privacy Policy.

How do I contact support?

Email support@relaykit.dev. We aim to reply within one business day. Include the model, request time and error message for API issues, or the transaction reference for billing questions.

Still stuck?

Read the docs and the error reference, or send us the details and we will help.