BestRelay pay-as-you-go with Claude Code for Sonnet 5.5 workdays

BestRelay lists Claude Sonnet 5.5 under its own name, and answers with it

BestRelay is an independent API relay, and the detail worth reading on its rate card is not a slogan but a name. When you send a request, BestRelay forwards it to the model provider under the model name you wrote. The model named is the model that answers, and nothing is quietly swapped for a cheaper tier behind the curtain. That rule is what the rest of the page rests on, and it is why a line-by-line reading of the Claude Sonnet 5.5 prices is in effect a reading of the service itself.

The line-up is curated. BestRelay carries 20 public models, each listed under its own name with four prices attached: input, output, cache write and cache read. Four numbers per model, no bundles, no averaged figure, no vague promise of provider rates. You can see what a call will cost before you make it, and you can see it in the same units the provider will later report back to you.

There is also a plain ending to the list. When a provider retires a model id, BestRelay delists it. A retired name disappears rather than lingering as a stub that routes somewhere unexpected. For a working day this matters less than the price does, but it matters for a related reason: a card is worth reading only if it can be believed. BestRelay keeps its catalogue tidy, its billing plain, and its promises to what the sheet can actually show.

The four lines of Claude Sonnet 5.5 on the BestRelay rate card

Take Claude Sonnet 5.5 and read the card one line at a time, with all prices given per 1,000,000 tokens. Input is $1.39. Output is $7.01. Cache write is $1.74. Cache read is $0.0944. Four lines, four separate ideas about what a call does: tokens arriving, tokens leaving, tokens written into the cache, and tokens fetched back out of it.

The official list price for the same model is printed alongside it: input $2, output $10, cache read $0.1. Cache write carries no official list price on the sheet, so BestRelay's $1.74 stands on its own line with nothing attached to it. Read the three comparable lines and the shape is plain. Input is $1.39 against $2, output is $7.01 against $10, and cache read is $0.0944 against $0.1, a fraction lower.

The neighbouring lines make the catalogue easier to place. Claude Sonnet 5 sits just below at $1.32 input, $6.67 output, $1.67 cache write and $0.1167 cache read. Claude Opus 5.5 sits above at $1.74, $8.13, $2.01 and $0.0778. Claude Opus 5 is $2.29, $11.46, $2.85 and $0.1389. Claude Haiku 4.5 is $0.4167, $2.01, $0.4861 and $0.0333. Claude Fable 5 is $4.17, $20.97, $5.14 and $0.3472, and Claude Fable 5.1 is $4.10, $20.42, $5.00 and $0.2083. BestRelay prints each of them under its own name, so choosing a model is a comparison you can make with your own eyes rather than a guess dressed up as advice.

A working day of Claude Sonnet 5.5, added up as an example

Here is a day of Claude Sonnet 5.5 work, worked through as an example rather than promised as an outcome. A day of 500,000 input tokens and 100,000 output tokens costs $1.40 on BestRelay. Twenty-two such days cost $30.71. Now let 350,000 of the input be read from cache, and the day costs $0.94. That is the shape of a real session: a long input context, a shorter answering stretch, and a good deal of repetition the cache can carry.

Set the same day against the official list price and the arithmetic is not subtle. The day is $2.00 at list, the 22 days are $44.00, and the cached day is $1.33. The difference on the single day is $0.60; the difference across the 22 days is $13.29. Over a working month, the gap is not a rounding error but a line in the budget, and it arrives without a subscription to cancel or a monthly fee to argue about.

Claude Sonnet 5, as a second example, behaves similarly: a day of 500,000 input and 100,000 output tokens costs $1.33 on BestRelay, 22 such days cost $29.18, and with 350,000 of the input read from cache the day costs $0.91. At the official list price the same day is $2.00, the 22 days $44.00 and the cached day $1.37, a difference of $0.67 on the day and $14.82 on the 22 days. Two models, two cards, the same method of reading them. BestRelay asks you to check the numbers yourself, which is the only sensible attitude toward a rate card.

For a smaller figure, GPT-6.1 Sol as an example: a day of 500,000 input and 100,000 output tokens costs $0.22 on BestRelay, 22 such days cost $4.74, and with 350,000 of the input read from cache the day costs $0.16. At the official list price the same day is $2.00, the 22 days $44.00 and the cached day $1.33, a difference of $1.78 on the day and $39.26 on the 22 days. The free credit alone covers about 1.6 million GPT-6.1 Sol input tokens.

BestRelay bills after the call, from the tokens the provider reports

Billing on BestRelay is prepaid and pay-as-you-go. There is no monthly fee, no subscription and no minimum spend. Each call is charged once, after it ends, from the tokens the provider reports, at the rate card prices. Nothing is held in advance, so a day of Claude Sonnet 5.5 costs what it costs and stops there.

The awkward cases are handled plainly. Requests the provider rejects are recorded at zero, and when the provider cannot be reached, nothing is charged. A failed call should not become a line of mysterious expenditure, and on BestRelay it does not.

Every new account opens with $0.35 of credit. Top-ups run through Stripe checkout in these amounts: $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89. Unused credit can be refunded on request within 13 days of the top-up, less what was consumed and any payment-processor fees; gift credit and commission cannot be refunded. Referrals pay 18% of the top-ups of people who joined through your link, and 4% of the top-ups of the people they invited, with commission movable into the balance or withdrawable to Alipay from $20.83. As an example, on an invited member's top-up of $13.89, the two commissions are $2.50 and $0.5556.

An account can hold several keys, each of which can be voided at any time, and every call appears in the usage records with its model, tokens and charge. When you want to know what a Tuesday cost, the records will say so, in the same units as the card.

Claude Code and Claude Desktop are the clients Claude models answer in

Claude models are served only to the official Claude Code and Claude Desktop clients, version 2.1.285 or newer. They do not work in Cursor, Cherry Studio, Cline, OpenCode, any SDK or cURL. Those tools, like Codex, pair only with GPT, DeepSeek, Kimi, GLM, Qwen and Grok. Gemini models use only their native protocol, through Gemini CLI or direct generateContent calls, and are not OpenAI-compatible. BestRelay states this in plain terms because a relay that pretends otherwise wastes an afternoon.

Connecting a client takes two settings: the address, and a key created under API keys in the dashboard. The details are in the BestRelay docs. One key allows 120 requests a minute, and one account allows 8 calls in flight. Keys can be voided whenever you like, and every call is recorded with its model, tokens and charge, so the running total never depends on your memory of the morning.

On privacy, BestRelay keeps the user-written text of requests for 30 days to review abuse and settle billing questions; system prompts and model replies are not stored. That is the whole retention story, stated in one line.

So the card reads as it should: named models, four prices each, a prepaid balance, no monthly fee, and a client rule that does not bend. If you would like to see how Claude Sonnet 5.5 behaves on your own work, BestRelay is ready when you are.