BestRelay API relay: naming GPT-6.1 Sol means GPT-6.1 Sol

The model id you write is the model that answers

You write a model id into your request. BestRelay forwards that request to the model provider under the name you wrote, so the model named is the model that answers, and nothing is quietly swapped for a cheaper tier on the way through. That is the whole job of an independent API relay, and BestRelay does it without ceremony: your name travels intact, and the reply comes back from the model you asked for.

The line-up is curated rather than sprawling. There are 20 public models, each listed under its own name, and each name carries four prices: input, output, cache write and cache read. When a provider retires a model id, BestRelay delists it. The list you read is therefore the list you can ask for, and a name that disappears does so because the provider withdrew it, not because the catalogue was tidied for convenience.

This matters on an ordinary afternoon at your own desk. You ask for Claude Opus 5.5 and Claude Opus 5.5 answers. You ask for GPT-6.1 Sol and GPT-6.1 Sol answers. If your workflow depends on the behaviour of a particular model, you can name it and then confirm it afterwards in the usage records, where every call appears with its model, its tokens and its charge. Nothing about the naming is a suggestion; it is the description of how the relay works.

BestRelay is an independent API relay, separate from Anthropic, OpenAI and Google. It sits between your client and the provider, and its task is to pass your request along faithfully and to bill you from what the provider reports. Naming a model on BestRelay is a small act of precision, and the rest of this article is a plain look at what follows it.

A sample day at your desk, shown as an example

Here is a sample day at your own desk, offered plainly as an example. Suppose you spend the day working with Claude Sonnet 5.5, writing 500,000 input tokens and 100,000 output tokens. On BestRelay that day costs $1.40, and 22 such days cost $30.71. If 350,000 of the input tokens are read from cache instead, the day costs $0.94. At the official list price the same day is $2.00, the 22 days $44.00 and the cached day $1.33, so the difference on the day is $0.60 and on the 22 days $13.29.

Now take a second example, with GPT-6.1 Sol. A day of 500,000 input tokens and 100,000 output tokens costs $0.22 on BestRelay, and 22 such days cost $4.74. 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; the difference on the day is $1.78 and on the 22 days $39.26.

Neither example asks you to commit anything in advance. You top up, you work, and the balance declines as the calls complete. The point of the figures is arithmetic rather than rhetoric: because the model you name is the model that answers, the tokens you can buy with a given balance are the ones the provider actually reports, priced at the rate card for that name. Cache reads are shown separately in both examples because they are charged separately on the card, and a day that reuses context is a day that costs less.

Both figures are examples, not quotes and not promises about your own usage. Your day will differ. What stays constant is the method: name the model, send the tokens, and read the charge back from the usage records.

Twenty curated models, each with four prices

The catalogue is short enough to read in one sitting. Claude Fable 5 is listed at $4.17 input, $20.97 output, $5.14 cache write and $0.3472 cache read per 1,000,000 tokens. Claude Fable 5.1 is $4.10 input, $20.42 output, $5.00 cache write and $0.2083 cache read. Claude Haiku 4.5 sits at $0.4167 input, $2.01 output, $0.4861 cache write and $0.0333 cache read. Claude Opus 5 is $2.29 input and $11.46 output, Claude Sonnet 5 is $1.32 input and $6.67 output, and Claude Sonnet 5.5 is $1.39 input and $7.01 output with a cache read of $0.0944.

The GPT names are listed the same way. GPT-5.6 Sol is $0.4167 input and $2.29 output, GPT-6 Astra is $0.7639 input and $3.96 output, and GPT-6.1 Sol is $0.2083 input and $1.11 output. The Gemini entries are Gemini 2.5 Flash at $0.2778 input and $2.08 output, Gemini 3.1 Pro at $1.74 input and $9.93 output, Gemini 3.8 Flash at $1.25 input and $5.56 output, and Gemini 4 Argon at $1.67 input and $8.06 output, each with cache write and cache read prices shown on the card.

The remaining names complete the twenty: DeepSeek V4 Pro is $1.53 input and $4.58 output with a cache read of $0.0514, Kimi K3 is $1.39 input and $7.01 output, Qwen3.8 Max is $0.8333 input and $2.57 output, Grok 4.6 and Grok 4.7 are $0.625 input, and GLM-5.3 is $0.625 input and $1.94 output. Because each model sits under its own name with its own four prices, choosing between them is a comparison you can make directly rather than an inference drawn from a single blended rate.

A prepaid balance, charged once per finished call

BestRelay runs on a prepaid pay-as-you-go balance. 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, and nothing is held in advance. If the provider rejects a request, the call is recorded at zero. If the provider cannot be reached, nothing is charged. The balance is a ledger of work already done rather than a commitment to work you might not do.

Every new account opens with $0.35 of credit, and the free credit alone covers about 1.6 million GPT-6.1 Sol input tokens. Top-ups run through Stripe checkout in fixed amounts: $1.39, $6.94, $13.89, $27.78, $69.44 or $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 are handled with the same plain arithmetic. You receive 18% of the top-ups of people who joined through your link, and 4% of the top-ups of the people they invited. Commission can be moved into the balance or withdrawn 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. Nothing about this arrangement changes the naming rule at the centre of BestRelay: the model you write is still the model that answers, and the charge still arrives once, after the call ends.

Clients, keys, limits, and the records you can read back

BestRelay serves Claude models only to the official Claude Code and Claude Desktop clients, version 2.1.285 or newer. Those models do not work in Cursor, Cherry Studio, Cline, OpenCode, any SDK or cURL. Tools such as Codex, Cursor, Cherry Studio, Cline, OpenCode, SDKs and cURL 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. Connecting a client takes two settings: the address and a key created under API keys in the dashboard, with the details kept in the docs.

An account can hold several keys, and each key can be voided at any time. One key is limited to 120 requests a minute, and one account to 8 calls in flight. Every call appears in the usage records with its model, its tokens and its charge, so the id you wrote is the id you can check afterwards, line by line, whenever you want to reconcile a day's spending.

On privacy, the text you write in requests is kept for 30 days so abuse can be reviewed and billing questions settled; system prompts and model replies are not stored. That is the whole arrangement: you name a model, the model answers, the tokens are counted by the provider, and the charge lands once in your records. If you would like to watch that happen at your own desk, the $0.35 opening credit is enough to send a starting request through BestRelay.