RelayVano
English
中文你好EnglishHelloEspañolHolaहिन्दीनमस्तेPortuguêsOláFrançaisBonjour
Log inSign up

← All postcards

RelayVano Pricing: What $0.40 Buys for Claude, GPT and Gemini

Fare watch EN

Six languages, one key, and a $0.40 starter balance

RelayVano is an independent relay made for developers worldwide. The whole site, the dashboard and the documentation come in six languages: 中文, English, Español, हिन्दी, Português, Français. Prices display in yuan on the Chinese pages and in approximate dollars on the others, so you can read the fares in the language and currency that suit you. New accounts start with $0.40 of credit, which is enough to send real requests, read the charge on the usage page and decide which model family deserves your next top-up.

One API key reaches every model through https://api.relayvano.com. The same key works with an OpenAI-compatible /v1 path, an Anthropic /v1/messages path and a native Gemini /v1beta path. Before you calculate anything, confirm the client you actually use, because the pairing is exact:

  • Claude models work only in the official Claude Code and Claude Desktop clients, version 2.1.285 or newer.
  • GPT, DeepSeek and the other non-Claude, non-Gemini models pair with Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL.
  • Gemini models use the native Gemini protocol only, in Gemini CLI. They are not OpenAI-compatible.

That short list is the first item on the checklist. A key that is right for Cursor is the wrong fit for a Claude Code workflow, and a Gemini request belongs in a native Gemini client rather than an OpenAI-compatible one.

What $0.40 buys in the Claude family

Claude fares are quoted per million tokens, and the charge itself is made in yuan. To make the starter credit meaningful, the examples below divide $0.40 by the prompt or reply fare and round to a friendly figure. They assume only one token type is billed, so a real conversation with both prompts and replies will consume the credit faster.

  • Claude Haiku 4.5 at $0.3958 prompt and $1.90 reply: about 1,010,000 prompt tokens, or roughly 210,000 reply tokens. The official list price is $1 / $5.
  • Claude Fable 5.1 at $4.07 prompt and $20.33 reply: about 98,000 prompt tokens, or roughly 19,000 reply tokens. The official list price is $10 / $50.
  • Claude Opus 5 at $1.88 prompt and $9.38 reply: about 212,000 prompt tokens, or roughly 42,000 reply tokens. The official list price is $5 / $25.
  • Claude Opus 5.5 at $1.65 prompt and $7.82 reply: about 242,000 prompt tokens, or roughly 51,000 reply tokens. The official list price is $4 / $20.
  • Claude Sonnet 5 at $1.27 prompt and $6.36 reply: about 314,000 prompt tokens, or roughly 62,000 reply tokens.
  • Claude Sonnet 5.5 at $1.25 prompt and $6.25 reply: about 320,000 prompt tokens, or roughly 64,000 reply tokens.

Two more items belong on a Claude checklist. First, cache hits are priced separately and cheaply, at $0.3125 for Fable 5.1, $0.0319 for Haiku 4.5, $0.1222 for Opus 5, $0.075 for Opus 5.5, $0.1111 for Sonnet 5 and $0.1097 for Sonnet 5.5, so a prompt-heavy workflow that reuses cached context stretches the starter credit further. Second, if a caller disconnects mid-stream, the request is still finished and billed on the tokens really used. Plan for that in your budget rather than assuming a broken connection refunds the work.

What $0.40 buys in the GPT and Gemini families

GPT models pair with Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL. Gemini models belong in Gemini CLI through the native Gemini protocol, so treat the two families as separate decisions rather than one combined pool.

  • GPT-6.1 Sol at $0.1958 prompt and $0.9792 reply: about 2,043,000 prompt tokens, or roughly 408,000 reply tokens. The official list price is $2 / $10.
  • GPT-5.6 Sol at $0.3736 prompt and $2.09 reply: about 1,070,000 prompt tokens, or roughly 191,000 reply tokens. The official list price is $5 / $30.
  • GPT-6 Astra at $0.7403 prompt and $3.70 reply: about 540,000 prompt tokens, or roughly 108,000 reply tokens. The official list price is $10 / $50.
  • Gemini 3.8 Flash at $1.20 prompt and $5.40 reply: about 333,000 prompt tokens, or roughly 74,000 reply tokens.
  • Gemini 3.1 Pro at $1.58 prompt and $9.13 reply: about 253,000 prompt tokens, or roughly 43,000 reply tokens.
  • Gemini 4 Argon at $1.58 prompt and $7.60 reply: about 253,000 prompt tokens, or roughly 52,000 reply tokens.

Cache-hit fares here are $0.0306 for GPT-5.6 Sol, $0.0736 for GPT-6 Astra, $0.0389 for GPT-6.1 Sol, $0.1194 for Gemini 3.8 Flash, $0.1569 for Gemini 3.1 Pro and $0.0792 for Gemini 4 Argon. If you mostly send long instructions and expect short answers, the prompt column matters more than the reply column. If your workload is generation-heavy, read the reply figures first. In every case, 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 your first $0.40 doubles as a live measurement of your own traffic pattern.

What $0.40 buys in DeepSeek and the other families

The remaining models work with Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL. Several of them stretch the starter credit a long way, which makes them a sensible first stop for exploring the relay.

  • DeepSeek V4 Flash at $0.3097 prompt and $1.24 reply: about 1,291,000 prompt tokens, or roughly 322,000 reply tokens.
  • DeepSeek V4 Pro at $1.37 prompt and $4.10 reply: about 291,000 prompt tokens, or roughly 97,000 reply tokens.
  • MiniMax M3 at $0.1472 prompt and $0.5889 reply: about 2,717,000 prompt tokens, or roughly 679,000 reply tokens.
  • Qwen3.7 Plus at $0.1458 prompt and $0.4361 reply: about 2,743,000 prompt tokens, or roughly 917,000 reply tokens.
  • Qwen3.8 Max at $0.7431 prompt and $2.23 reply: about 538,000 prompt tokens, or roughly 179,000 reply tokens.
  • Kimi K3 at $1.25 prompt and $6.24 reply: about 320,000 prompt tokens, or roughly 64,000 reply tokens.
  • Hunyuan 4 at $0.3514 prompt and $1.12 reply: about 1,138,000 prompt tokens, or roughly 357,000 reply tokens.
  • Grok 4.6 at $0.575 prompt and $1.72 reply: about 695,000 prompt tokens, or roughly 232,000 reply tokens.
  • Grok 4.7 at $0.6194 prompt and $1.86 reply: about 645,000 prompt tokens, or roughly 215,000 reply tokens.
  • MiMo V2.6 Pro at $0.2319 prompt and $0.4653 reply: about 1,724,000 prompt tokens, or roughly 859,000 reply tokens.

MiMo V2.6 Pro is worth a second look for repeated context, with a cache hit at $0.002778, while Qwen3.8 Max sits at $0.0889, Kimi K3 at $0.125, Grok 4.6 at $0.1431, Grok 4.7 at $0.1542, MiniMax M3 at $0.0292, Qwen3.7 Plus at $0.0431, Hunyuan 4 at $0.0167, DeepSeek V4 Flash at $0.008333 and DeepSeek V4 Pro at $0.0458. A quick test on one of these families can tell you how much of your own traffic is prompt, reply or cache before you commit to a larger balance.

Before you sign up: five checks that protect your budget

RelayVano sells nothing but pay-as-you-go balance. There is no monthly fee, you pay per token, and the balance never expires. Here is the checklist to run before you sign up:

  • Confirm the client pairing. Claude Code or Claude Desktop for Claude, Gemini CLI for Gemini, and the OpenAI-compatible clients and SDKs for the rest.
  • Estimate your mix of prompt tokens, reply tokens, cache stores and cache hits, then divide $0.40 by the matching fare for your chosen family.
  • Check the usage page after your first requests. Each one lists its charge, so a misconfigured client shows up immediately.
  • Know how you will top up. 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.
  • Decide whether to invite others. Referrals return 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 into the balance.

One more practical note: if something does not look right, logged-in members write to us under Messages in the dashboard. There is no need to hunt for a separate support channel.

Start with the $0.40 credit, send a handful of requests from the client you already use, and read the usage page. You will quickly see which model family fits your traffic and how far the balance stretches. When you are ready, sign up for RelayVano, top up through Stripe checkout, and keep every model you need behind one key and one prepaid balance.

More postcards from this route