Grok 4.7 pricing and practical tips for every Fluxio top-up

§1 Match a pay-per-token top-up to your bot's real rhythm

For a hobby bot, the trick is to know how fast your balance drains before you commit a large amount to it. Fluxio's pay-per-token balance has no monthly fee, and top-ups come in six sizes: $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89, paid through Stripe Checkout. A practical habit is to begin with a smaller top-up, watch the usage log for a day or two, and then move up a tier once you know the burn rate of the model your bot actually calls.

• – Every new account receives $0.12 in sign-up credit, which is enough to cover a handful of small test calls while you tune prompts and parameters.
• – Unused balance from a top-up can be refunded within 4 days of that top-up, minus what was consumed and payment fees. You ask through the message page.
• – Referrals return 17% of what a directly invited friend pays, and 5% for the next level. Rebates can move into your balance or be withdrawn from $12.50 by Alipay.
• – Fluxio's usage log records each request, its tokens and its cost, so you can trace exactly where a top-up went.

Because refunds are only available inside a 4-day window on unused balance, keep early experiments small and deliberate. If your bot is still being rewritten, a $1.39 or $6.94 top-up gives you room to test without tying up funds you may want back.

§2 Read Grok 4.7 pricing as a per-request budget

Fluxio bills on real usage, and each request is charged on the tokens the model reports, with four token types priced separately: input, output, cache write and cache read. There is no flat per-request fee. For Grok 4.7, the per 1M token prices are $0.6069 input, $1.82 output and $0.1514 cache read, so output is the line item that usually dominates a chatty bot.

Here is a worked example. If your bot sends 100 messages a day, and each message carries 1,000 input tokens and 500 output tokens, that is 100,000 input tokens and 50,000 output tokens per day. At Grok 4.7 pricing, input costs $0.06069 and output costs $0.091 per day, or about $0.15169 per day and roughly $4.55 over 30 days. That is an example, but it shows why request shape matters more than headline rates: trimming output tokens usually saves more than trimming input.

Different loops suit different price points. For quick utility bots, MiniMax M3 is priced at $0.1514 input and $0.6028 output per 1M tokens, MiMo V2.6 Pro at $0.2111 input and $0.4208 output, and GPT-6.1 Sol at $0.1944 input and $0.9722 output. For longer reasoning work, GPT-6 Astra is $0.6806 input and $3.40 output, Qwen3.8 Max is $0.7528 input and $2.26 output, and Grok 4.6 is $0.5639 input and $1.69 output. DeepSeek V4 Flash sits at $0.2875 input and $1.15 output with cache read at $0.006944 per 1M tokens, while Hunyuan 4 is $0.3361 input and $1.08 output with cache read at $0.0167.

Some Fluxio rates sit well under the published list prices. Claude Fable 5 is $3.50 input and $17.50 output per 1M tokens against an official list price of $10.00 input and $50.00 output, and Claude Haiku 4.5 is $0.3458 input and $1.66 output against a list price of $1.00 input and $5.00 output. GPT-5.6 Sol is $0.3861 input and $2.16 output against a list price of $4.00 input and $20.00 output, with the list price described as an OpenAI promotional price announced until at least 2026-11-21.

§3 Use the client that actually matches your model

Client compatibility is where hobby builds often stall, and the rules in Fluxio are specific rather than universal.

• – Claude models work only in the official Claude Code and Claude Desktop apps, version 2.1.285 or newer. They do not work in Cursor, Cherry Studio, Cline, OpenCode, any SDK or cURL.
• – Gemini models work only through their native protocol, which means Gemini CLI. They are not OpenAI-compatible, so they do not run in OpenAI-format clients.
• – GPT, DeepSeek, Kimi, GLM, Grok, Qwen, Hunyuan, MiniMax and MiMo work from any OpenAI-format client, including Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL. Codex runs GPT models.
• – The API base is api.fluxioapi.com, and OpenAI-style clients add /v1 to it.

So if your hobby bot is built in Cursor or Cline, Grok 4.7, GPT-6.1 Sol, DeepSeek V4 Flash and the rest of the OpenAI-format family are available to you. If you want to work with Claude models, set up Claude Code or Claude Desktop instead. If you want Gemini, use Gemini CLI. Picking the right client first saves tokens that would otherwise be spent on failed calls, and it keeps your Fluxio usage log clean so the cost of each working request is easy to read.

§4 Let streaming show you where your tokens go

Fluxio is streaming-first. Replies arrive token by token over server-sent events on the OpenAI, Anthropic and Gemini protocols, and every stream reports its real usage. That visibility is the practical lever for stretching a top-up: you can watch a reply grow and judge whether an expensive output is justified before the request finishes.

Dropped connections are handled in a specific way. If your client disconnects mid-stream, Fluxio still reads the reply to its end and bills the tokens actually used, nothing more. It never adds a penalty. Tokens already generated are not refunded, so the practical lesson is to keep your client connected and let the stream complete.

Cache pricing is where careful bots quietly save. Cache reads are charged separately from cache writes, and the spread between them varies by model. Claude Fable 5 charges $0.2694 for cache read and $4.31 for cache write per 1M tokens, while Claude Fable 5.1 charges $0.2319 and $4.78. Claude Sonnet 5.5 is $0.0917 for cache read and $1.49 for cache write, and Claude Opus 5.5 is $0.0681 and $1.78. On the Gemini side, Gemini 2.5 Flash is $0.0208 for cache read and $0.2097 for cache write, and Gemini 4 Argon is $0.0736 and $1.47. If your bot reuses the same long context repeatedly, comparing cache read and cache write rates before you commit a top-up can change your effective cost more than switching between close models.

§5 Keep limits, logs and plan allowances in view

Fluxio allows 120 requests a minute per key and 8 streams open at once per account. For a hobby bot that runs a few concurrent tasks, that is comfortable room, but a retry loop that fans out across many requests can hit the per-key limit quickly. Watching the usage log helps you spot runaway loops, and the interface is available in six languages: English, Chinese, Spanish, Hindi, Portuguese and French.

When a top-up is not the right shape for your usage, monthly plans are available, and their allowances are counted at the plan rate rather than list prices, with a multiplier of × 2.8. The Claude monthly plan at $62.36 gives an allowance of up to $1100 over 30 days, at most $350 a week, for Claude models only with your own API key. The $165.14 Claude monthly plan allows up to $2900 over 30 days, at most $850 a week, on the same terms. The GPT monthly plan at $59.58 covers up to $2600 over 30 days, at most $800 a week, for GPT models only with your own API key, and the $183.19 plan allows up to $8000 over 30 days, at most $2400 a week.

Support requests go from the site contact page or the dashboard, and replies appear in the dashboard. There is no support email to wait on. Keeping your usage log, your stream limits and your remaining balance visible in one place is what lets a small hobby bot keep running on a top-up you sized with actual numbers rather than guesses.