BestRelay with GPT-6.1 Sol: an API relay priced per token

BestRelay forwards each request under the model name you wrote

BestRelay is an independent API relay, and its line-up is curated rather than infinite: 20 public models, each listed under its own name with four prices — input, output, cache write and cache read. When a provider retires a model id, BestRelay delists it, so the list you read is a list of models that can still answer.

The point of that care is simple. BestRelay forwards each request to the model provider under the model name the caller wrote, so the model named is the model that answers, and nothing is quietly swapped for a cheaper tier. If you ask for GPT-6.1 Sol, GPT-6.1 Sol is what replies. There is no second, less visible model doing the work under a familiar label.

That sounds like a small promise, but in daily use it is the one that decides whether a desk tool feels dependable. You choose a model for a reason — a writing task, a long refactor, a quick classification — and you should not have to audit the plumbing each time to confirm you got it. BestRelay's answer is structural rather than rhetorical: the name in the request is the name on the receipt, and the usage records show the model for every call.

The rest of the service follows the same plainness. A prepaid pay-as-you-go balance, no monthly fee, no subscription, 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. Requests the provider rejects are recorded at zero, and when the provider cannot be reached, nothing is charged.

Pointing Codex at BestRelay with GPT-6.1 Sol as the default model

Here is a sample day at your own desk, plainly labelled as an example. You keep Codex open and set GPT-6.1 Sol as the default model for the session. Codex, like Cursor, Cherry Studio, Cline and OpenCode, pairs with the GPT, DeepSeek, Kimi, GLM, Qwen and Grok families on BestRelay; the Claude models are served only to the official Claude Code and Claude Desktop clients, version 2.1.285 or newer, and are not available in Codex.

Because the name you write is the name that answers, setting the default once is enough. You can change it when a different job calls for a different temperament, and the switch is visible in the usage records rather than buried in a dashboard you have to interpret afterwards.

For example, a day of 500,000 input and 100,000 output tokens costs $0.22 on BestRelay, 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 $1.48; the difference on the day is $1.78 and on the 22 days $39.26.

The per-token card behind that day reads: GPT-6.1 Sol at input $0.2083, output $1.11 and cache read $0.0431 per 1,000,000 tokens, against an official list price of input $2, output $10 and cache read $0.5.

For example, the $0.35 of credit that opens a new account covers about 1.6 million GPT-6.1 Sol input tokens on its own — enough to watch the day above run without reaching for a card at all.

A prepaid balance, charged once per call and nothing held in advance

BestRelay bills from a prepaid pay-as-you-go balance. There is no monthly fee, no subscription and no minimum spend; you are not committing to a tier before you have measured anything. Each call is charged once, after it ends, using the token counts the provider reports and the rate card prices for the model you named.

Failure is handled with the same economy. Requests the provider rejects are recorded at zero, and when the provider cannot be reached, nothing is charged. A failed attempt does not quietly become a bill, which matters on a day when a tool is making many small requests and you would rather read one honest total than a row of surprises.

Top-ups run through Stripe checkout in fixed amounts: $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89. For example, if you choose $13.89, that is the whole transaction — no monthly charge waiting behind it. 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.

The referral scheme is modest and readable: 18% of the top-ups of people who joined through your link, and 4% of the top-ups of the people they invited. For example, on an invited member's top-up of $13.89, the two commissions are $2.50 and $0.5556. Commission can be moved into the balance, or withdrawn to Alipay from $20.83.

Twenty curated models, each with its own four prices

The line-up is 20 public models, and each is listed under its own name with four prices: input, output, cache write and cache read. You can read across the card without guessing what a model will cost to run. GPT-6.1 Sol sits at input $0.2083, output $1.11 and cache read $0.0431 per 1,000,000 tokens; GPT-5.6 Sol at input $0.4167, output $2.29 and cache read $0.0333; GPT-6 Astra at input $0.7639, output $3.96 and cache read $0.0792.

Claude Sonnet 5.5 is priced at input $1.39, output $7.01, cache write $1.74 and cache read $0.1222. For example, a day of 500,000 input and 100,000 output tokens costs $1.40 on BestRelay, 22 such days cost $30.71; with 350,000 of the input read from cache the day costs $0.95. At the official list price the same day is $3.00, the 22 days $66.00 and the cached day $2.06; the difference on the day is $1.60 and on the 22 days $35.29.

Two client rules are worth keeping in mind. 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. Gemini models use only their native protocol, through Gemini CLI or direct generateContent calls, and are not OpenAI-compatible. Gemini 3.1 Pro, for its part, is listed at input $1.74, output $9.93, cache write $1.74 and cache read $0.1389.

Two settings, a dashboard, and a record of every call

Connecting a client takes two settings: the address and a key created under API keys in the dashboard. The details live in the docs, and the address itself is yours to keep — it does not belong in an article. Create the key, paste the pair into the client, and the model list opens up.

The limits are stated plainly: 120 requests a minute on one key, and 8 calls in flight on one account. An account can hold several keys, each can be voided at any time, and every call appears in the usage records with its model, tokens and charge. If a key travels further than it should, you can end it and see what it did.

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. The record is therefore about tokens and charges, not a second archive of what you typed.

None of this asks for a long commitment. There is a prepaid balance, a card of named models and prices, and a client that talks to the model you chose. Point Codex at BestRelay with GPT-6.1 Sol as your default and see how the day reads on the usage records.