RelayVano API Relay Pricing: Claude and Gemini in Six Languages

Yuan on Chinese pages, dollars on the rest

Most people see a dollar sign next to a model price and quietly assume three things: the number is exact, the charge will arrive in dollars, and the same figure will appear no matter where you sign in. RelayVano is more straightforward about what actually happens. Prices show in yuan on the Chinese pages and in approximate dollars on the others. The charge itself is made in yuan, and the dollar figures you read in English, Español, हिन्दी, Português and Français are conversions for orientation, with the reference rate at 1 USD ≈ ¥7.2.

That distinction matters when you compare a RelayVano fare with a vendor subscription or with a published list price in dollars. You are reading the same underlying price either way, just presented in the currency that fits the page you opened. Nothing is hidden behind a regional toggle or a second pricing table. The whole site, dashboard and docs come in six languages: 中文, English, Español, हिन्दी, Português, Français.

So if you land on the Chinese version and think the yuan figure is a local-only arrangement, or if you land on the English version and think the dollar figure is what your card will be charged, both assumptions are wrong. The number is the price. The currency of settlement is the yuan.

One key, three protocols, and exact client pairings

A common assumption about any relay service is that one key works with every tool in your editor. With RelayVano, one API key does reach every model through https://api.relayvano.com, but the client pairing is specific, and there are no exceptions.

Claude models work only in the official Claude Code and Claude Desktop clients, 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. Gemini models are not OpenAI-compatible, and it is worth saying plainly so nobody spends an afternoon debugging a request format that was never going to be accepted.

Underneath, RelayVano speaks three familiar shapes: an OpenAI-compatible /v1 route, Anthropic /v1/messages, and native Gemini /v1beta. Setup is a matter of pointing your client at the base address and supplying your key. If you work across model families, the practical question is not whether a key exists but which client belongs to which model, and RelayVano keeps that mapping narrow enough to remember.

What a headline price does not tell you

Another assumption is that a per-model fare is a single number covering the whole exchange. RelayVano bills at token level. 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 you can tie a line item back to the request that produced it.

RelayVano is prepaid. You pay per token, there is no monthly fee, and the balance never expires. That last point quietly removes a familiar worry: a small balance left over after a busy week is still there months later, still spendable, still visible in the dashboard.

Streaming behaves in a way that many readers will appreciate. If a caller disconnects mid-stream, the request is still finished and billed on the tokens really used. You are not paying for a response you never saw, and you are not being billed for work the provider did not complete either. The usage page remains the place to confirm what a given request cost.

Reading the fares against the official list prices

RelayVano publishes approximate US dollar fares per million tokens, and for several models the comparison with the official list price is visible at a glance. Claude Fable 5.1 is $4.07 prompt / $20.33 reply / $0.3125 cache hit, against an official list price of $10 / $50. GPT-6.1 Sol is $0.1958 prompt / $0.9792 reply / $0.0389 cache hit, against an official list price of $2 / $10. Gemini 3.1 Pro is $1.58 prompt / $9.13 reply / $0.1569 cache hit. DeepSeek V4 Flash is $0.3097 prompt / $1.24 reply / $0.008333 cache hit.

Here is an example, not a forecast. Suppose you send 200 requests a day, and each request averages 500 prompt tokens and 500 reply tokens on MiniMax M3, priced at $0.1472 prompt and $0.5889 reply per million tokens. One request then costs roughly $0.0736 plus $0.29445, or about $0.368. At 200 requests, that example comes to about $73.61 a day. Change the model, the token counts or the request volume and the sum changes; the point is that you can work it out from published fares instead of guessing.

The same exercise works for a heavier model. GPT-6.1 Sol at 1,000 prompt and 1,000 reply tokens per request is about $1.175 per request, so 200 such requests a day in this example would be about $235. RelayVano gives you the figures; the arithmetic is yours.

Top-ups, refunds, referrals and a real support channel

Top-ups go through Stripe checkout. Cards work everywhere, and WeChat Pay is available where your account supports it. Unspent top-ups can be refunded within 9 days, which makes a first experiment a low-commitment one. New accounts start with $0.40 of credit, enough to make a small request and see the whole flow, including the usage page, before you add funds.

The referral programme is also simple to describe. You earn 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 and spent on tokens.

Support is not a guessing game either. Logged-in members write to us under Messages in the dashboard, where your account and your usage history are already attached to the conversation.

Made for developers worldwide, in six languages

RelayVano is an independent relay made for developers worldwide, and the six-language site, dashboard and docs are part of that promise rather than a decorative extra. Read the fares in yuan on the Chinese pages, in approximate dollars elsewhere, and use the same key and the same base address either way. Choose the client that matches your model: Claude Code or Claude Desktop for Claude, Gemini CLI for Gemini, and Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs or cURL for GPT, DeepSeek and the rest.

There is no monthly fee, no expiring balance and no per-request charge to forget. You top up, you use tokens, you watch the usage page. If that sounds like the relay you have been looking for, create your account, collect the $0.40 starter credit, and send your first request through RelayVano.