RelayVano Pay-as-you-Go Setup: Disable Claude Code Keys in a Click
Why one RelayVano key per project keeps your work tidy
Every project you run tends to bring its own habits: a different client, a different model family, a different monthly traffic pattern, a different teammate who needs access. The easiest way to keep those things apart is to give each project its own RelayVano API key. One key for the Claude Code work, one for a Codex experiment, one for a Gemini CLI script, and one for anything you are still trying out. Because every key is a separate object in the dashboard, you can see at a glance which project is active, and you can switch one off without touching the rest.
That separation matters most on the days when something goes sideways. If a script starts looping, if a teammate leaves a job running overnight, or if you simply want to pause a side experiment for a week, you do not have to hunt through shared configuration files or rebuild credentials. You open the dashboard, find the key that belongs to that project, and disable it in a click. The other keys keep working exactly as before.
RelayVano is an independent relay made for developers worldwide, and the whole site, dashboard and docs come in six languages: 中文, English, Español, हिन्दी, Português, Français. Prices show in yuan on the Chinese pages and in approximate dollars on the others, so a team working across several regions can read the same numbers in a language they are comfortable with.
Setting up one key per project in the dashboard
Creating the keys takes a few moments. Sign in to RelayVano, open the API key area of the dashboard, and create a new key. Give it a name that matches the project rather than something vague: a name like a project folder, a client, or a short ticket reference will save you guessing later. Copy the key once, then paste it into that project's environment settings, where your client looks for the base address and the key value.
All traffic goes through a single base address, https://api.relayvano.com, and the pairing rules are specific so that nothing fails quietly:
- Claude models work only in the official Claude Code and Claude Desktop clients, at version 2.1.285 or newer.
- Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL pair only with GPT, DeepSeek and the other non-Claude, non-Gemini models.
- Gemini models use the native Gemini protocol only, in Gemini CLI. They are not OpenAI-compatible, so pointing an OpenAI-style client at them will not work.
With those rules in place, label your keys by client as well as by project. A key named for a Claude Code project and a key named for a Codex project make it obvious which one belongs where, and if you ever misconfigure a client, you know exactly which key to look at first.
Worked example: what a Claude Code project costs per month
Here is a worked example, and every figure below is an example rather than a promise about your own usage. Suppose a Claude Code project sends 200 requests a day. Each request uses 1,000 prompt tokens, 500 reply tokens, and 2,000 cache hit tokens. We price the request with Claude Sonnet 5.5, whose approximate fares are $1.25 per million prompt tokens, $6.25 per million reply tokens, and $0.0889 per million cache hit tokens.
First the prompt tokens: 1,000 divided by 1,000,000 is 0.001 million tokens, and 0.001 multiplied by $1.25 gives $0.00125 per request. Next the reply tokens: 500 divided by 1,000,000 is 0.0005 million, and 0.0005 multiplied by $6.25 gives $0.003125. Then the cache hits: 2,000 divided by 1,000,000 is 0.002 million, and 0.002 multiplied by $0.0889 gives $0.0001778. Adding the three parts, $0.00125 plus $0.003125 plus $0.0001778, gives $0.0045528 per request.
Now scale it up. Two hundred requests a day gives $0.91056 for the day, because $0.0045528 multiplied by 200 is $0.91056. Over 30 days that is $27.3168, because $0.91056 multiplied by 30 is $27.3168. That is the whole monthly shape of this example project, and it is why one key per project is useful: you can point at the key and ask whether the traffic still looks like the traffic you planned for.
RelayVano bills against a prepaid balance, pay per token, with no monthly fee, and the balance never expires. Prompt tokens, reply tokens, cache stores and cache hits are priced separately, and every request shows up with its charge on the usage page, so the sum above can be checked line by line rather than taken on trust.
Worked example: a lighter project on a different key
Now take a second example project, again 200 requests a day with 1,000 prompt tokens, 500 reply tokens and 2,000 cache hit tokens per request, but running on MiniMax M3 instead. The approximate fares are $0.1472 per million prompt tokens, $0.5889 per million reply tokens, and $0.0292 per million cache hit tokens.
The prompt part is 0.001 multiplied by $0.1472, which is $0.0001472. The reply part is 0.0005 multiplied by $0.5889, which is $0.00029445. The cache hit part is 0.002 multiplied by $0.0292, which is $0.0000584. Together that is $0.00050005 per request. Two hundred requests a day comes to $0.10001, and 30 days of that comes to $3.0003.
Two keys, two clearly different monthly figures, and no need to merge them into a single opaque number. If the lighter project is the one you want to pause, you disable that key and leave the Claude Code project untouched. New accounts start with $0.40 of credit, which is enough to make the first request and see how the usage page records it.
Disabling a key without disturbing the rest of your setup
Disabling is deliberately a small action. Open your key list in the RelayVano dashboard, find the key attached to the project you want to stop, and click to disable it. Requests that try to use that key afterwards stop being accepted, while every other project key continues to work normally. If a scheduled job fails because its key was switched off, the fix is one click away rather than a debugging session.
Turning a key off does not erase the history of what it did. The usage page keeps showing each request with its charge, so you can review the traffic that led to the decision. If a caller disconnects mid-stream, the request is still finished and billed on the tokens really used, which means a half-finished call cannot quietly disappear from the record.
Disabling is also reversible in the way you would want. If a project comes back to life next week, you can re-enable the same key instead of creating a new one and updating your environment settings again. Keeping one key per project means your configuration stays stable, and your options stay open.
Paying, topping up and getting help
RelayVano sells nothing but pay-as-you-go balance. Top-ups go through Stripe checkout, with cards available everywhere, plus WeChat Pay where your account supports it. Unspent top-ups can be refunded within 9 days, so a large balance does not have to feel like a commitment.
There is also a way to earn from the service you already use. Referrals pay 16% of every top-up by people who sign up through your link, and 4% one level further. Earnings can be withdrawn from $5.56 or moved straight into your balance.
If something is unclear, logged-in members write to us under Messages in the dashboard, where your project details and key history are already in front of you.
One key per project, one click to disable, every charge visible: that is the whole model. Create your RelayVano account, add $0.40 of starter credit, and set up your first project key today.