As of October 8, 2026, Relaykit has a new customer console at /console. It replaces the dashboard we used before. Everything a customer does after signing up now happens in one place, in the same design as the rest of the site: creating and limiting API keys, seeing what each request cost, and adding credit.
If you already had an account, nothing moves. Sign in with the same credentials. Your keys, balance and history are all there.
What is in it
The console has five sections:
| Section | What you do there |
|---|---|
| Overview | See your balance and recent spend, plus the base URLs to paste into your code |
| API keys | Create, limit, rename and delete keys |
| Usage | A log of every request: model, key, tokens and cost |
| Billing | Your balance, and where you add credit |
| Settings | Account details and password |
API keys with spend limits and expiry dates
Keys are where most of the change is. When you create a key, you can now set:
- A name. It shows up next to every request in the usage log. Name keys after the project and environment, for example
invoice-bot-prod. Then “what does this cost?” takes one filter to answer. - A spend limit. The most this key can ever spend. When the key reaches it, its requests are rejected with HTTP
403before they reach a model, so the blocked request costs nothing. Your other keys keep working. The limit counts down as you use the key and does not reset by itself. Raise it when you decide to. - An expiry date (optional). After that date the key returns
401. Use it for demos, contractors and short experiments, so the key is safe even if nobody remembers to delete it.
Every key spends from the same account balance, and a request needs room in both the key’s limit and the balance. That means one runaway script can only use its own key’s limit, not everything you have loaded.
Deleting a key takes effect on the next request that uses it. Its past requests stay in the usage log under its name, so you keep the history for that project even after the key is gone.
We wrote a tutorial on how to split keys and size each limit from real traffic: give every project its own API key and spend limit.
A usage log you can answer questions with
The usage page lists every request: when it happened, which key made it, which model ran, the input and output tokens, and what it cost. You can filter by key, model and date range. The tokens are the same numbers returned in each API response’s usage object, priced at the rates on the models page.
Some questions it answers directly:
- Which project, or which developer’s coding agent, spent the most this week?
- Did yesterday’s deploy make the average request longer?
- Was the overnight spike one key or all of them?
A request rejected before it reaches a model, for example a 401 or a 403 from a spend limit, is not billed.
Balance and adding credit
The billing page shows your current balance in US dollars. Each request’s cost is deducted from it at the published per-token rates. It is a prepaid balance: when it reaches zero, requests stop. You are never billed afterwards, so the most you can spend is what you have loaded.
You add credit on the billing page. It shows only the payment options that work today. Card checkout for the credit packs listed on the pricing page is not live yet. We will post a dated update here when it is.
New accounts start at $0, with no sign-up credit. The smallest top-up is $3, and models become available once the account has paid $3 or more.
Base URLs on the overview page
The overview page shows the two base URLs you need, ready to copy:
{SITE.apiBase}/v1for OpenAI-format clients (the OpenAI SDKs, Codex CLI, Cursor, Cline and most other tools){SITE.apiBase}for Anthropic-format clients (the Anthropic SDKs and Claude Code)
The difference is whether you include /v1, and it causes most first-request errors. The quickstart walks through a first call in both formats.
What we did not change
- Rates. Per-token prices are the same, and every rate is on the models page.
- The API. Endpoints, model IDs and request formats are unchanged. Existing keys keep working without any change to your code.
- Where your requests go. The console is a new interface on the same account system. It doesn’t change how requests are routed or metered.
Known gaps
We’d rather list these than have you discover them:
- Card payments are not available yet (see above).
- Teams. Each account has one login. To share spend across a team today, give each person their own key on the shared account. Shared logins with roles are something we want to build, but there is no date yet.
- Usage export to CSV is not in this release. The log is on screen only for now.
If something in the console is unclear or broken, email support@relaykit.dev and include the time of the request and the key name. Never include the key itself.