RelayVano Pay-as-You-Go API Relay: Claude Sonnet 5 and GPT-6.1 Sol

What a Small Team Assumes About Separate Official Accounts

When a small team plans its AI budget, a familiar assumption is that each family of models needs its own official account, its own billing page and its own monthly subscription. Three vendors, three sign-ups, three invoices, and a spreadsheet to work out who spent what. Add the cost of unused seats and the surprise of a charge that arrives after a busy week, and the picture looks predictable enough to sign off on without much thought.

That assumption is worth testing against how RelayVano actually behaves. We are 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 time zones is not forced to read a pricing page in a language that is not its own.

RelayVano sells nothing but pay-as-you-go balance. There is no monthly fee, the balance never expires, and you pay per token for the work your code actually does. New accounts start with $0.40 of credit, which is enough to make a first request and see exactly how the usage page works before you commit any larger amount.

One Key Instead of a Stack of Vendor Logins

The practical difference shows up in the setup. One API key reaches every model through https://api.relayvano.com, using the OpenAI-compatible /v1 route, the Anthropic /v1/messages route and native Gemini /v1beta. For a team, that means one credential to distribute, one balance to watch and one place to look when a developer asks why a request cost what it cost.

There is one pairing detail to respect, and it is a clean one rather than a vague one. Claude models work only in the official Claude Code and Claude Desktop clients, version 2.1.285 or newer. Gemini models use the native Gemini protocol only, in Gemini CLI, and they are not OpenAI-compatible. Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL pair only with GPT, DeepSeek and the other non-Claude, non-Gemini models.

In other words, you keep the tools your developers already prefer, and you point the right one at the right family of models. The setup is a matter of choosing the matching route in your client configuration and adding your RelayVano key. There is no need to invent a workaround for a client that was never designed to speak a given protocol.

Claude Sonnet 5 Pricing, Compared Line by Line

Let us take one model and compare the numbers instead of arguing from instinct. RelayVano lists Claude Sonnet 5 at $1.27 prompt, $6.36 reply and $0.1111 cache hit per million tokens, against an official list price of $2 / $10 / $0.2 cache hit. The charge itself is made in yuan, with fares quoted here at approximately 1 USD ≈ ¥7.2.

As an example, suppose your team sends 200 requests a day, and each request averages 1,000 prompt tokens and 500 reply tokens. That is 200,000 prompt tokens and 100,000 reply tokens per day. On RelayVano, the prompt side comes to about $0.254 and the reply side to about $0.636, or roughly $0.89 per day, or about $26.70 over 30 days. At the official list price, the same example works out to $0.40 plus $1.00, which is $1.40 per day and $42.00 over 30 days.

Cache hits are priced separately, and they are worth watching in any real workload. A prompt that reuses a cached prefix at $0.1111 per million tokens on RelayVano is priced well below the $0.2 official list figure for the same category. Prompt tokens, reply tokens, cache stores and cache hits each carry their own rate, so a team can tune its own usage rather than guessing which line item is doing the work.

GPT and Gemini Without a Second Budget

The same example applied to GPT-6.1 Sol gives another useful reference point. RelayVano lists $0.1958 prompt, $0.9792 reply and $0.0389 cache hit per million tokens, against an official list price of $2 / $10 / $0.1 cache hit. Under the 200-requests-a-day example above, that is about $0.039 for prompts plus $0.098 for replies, roughly $0.137 per day and about $4.11 over 30 days, compared with $42.00 at the official list price for the same shape of traffic.

For Gemini, RelayVano lists Gemini 3.1 Pro at $1.58 prompt, $9.13 reply and $0.1569 cache hit per million tokens. Gemini 3.8 Flash is listed at $1.20 prompt, $5.40 reply and $0.1194 cache hit, and Gemini 4 Argon at $1.58 prompt, $7.60 reply and $0.0792 cache hit. Because Gemini models run through the native Gemini protocol in Gemini CLI, a team can bring a different model family into the same account and the same prepaid balance without adding another vendor relationship to manage.

There is room for the lighter options too. MiniMax M3 is listed at $0.1472 prompt and $0.5889 reply, Qwen3.7 Plus at $0.1458 prompt and $0.4361 reply, and MiMo V2.6 Pro at $0.2319 prompt and $0.4653 reply per million tokens. DeepSeek V4 Flash is listed at $0.3097 prompt and $1.24 reply, and DeepSeek V4 Pro at $1.37 prompt and $4.10 reply. A small team can match each task to a fare rather than paying one rate for everything.

How the Balance Behaves When Something Goes Wrong

Budget tools tend to be judged by what happens in the awkward moments, and here the behaviour is straightforward. If a caller disconnects mid-stream, the request is still finished and billed on the tokens really used. Nothing is quietly written off, and nothing is charged for tokens that were never produced.

Every request shows up with its charge on the usage page, which turns the budget conversation into a list a developer can actually read. Top-ups go through Stripe checkout, with cards accepted everywhere, plus WeChat Pay where your account supports it. Unspent top-ups can be refunded within 9 days, and because the balance never expires, an unused top-up does not quietly disappear on a renewal date.

Support is also part of the same place. Logged-in members write to us under Messages in the dashboard, so a question about a charge, a client pairing or a refund does not have to become a separate support channel. If you are wondering whether a particular model fits a particular client, ask before you wire it in.

Sharing the Account Without Sharing the Bill

There is another assumption that separate official accounts encourage: that growth has to mean more subscriptions. With RelayVano, a team grows on the same prepaid balance, topped up when it makes sense for the team. Referrals can help that balance stretch further. People who sign up through your link contribute 16% of every top-up, and 4% one level further, with earnings that can be withdrawn from $5.56 or moved into the balance.

So the honest comparison is not one RelayVano subscription against another vendor's subscription. It is a single prepaid, pay-as-you-go balance with no monthly fee and no expiring credit, set against several official accounts, several billing cycles and several places to check usage. For a small team, the difference is less about the headline rate and more about the number of moving parts you stop carrying.

RelayVano gives you one key, one dashboard, one usage page and fares you can read in your own language, with Claude, GPT, Gemini, DeepSeek and other models reachable through the clients that support them. Create an account, start with the $0.40 of new-account credit, send a request and read the charge on the usage page. Then decide whether your budget still needs three separate official accounts. Sign up for RelayVano and put the balance to work.