BestRelay and DeepSeek V4 Pro: pricing on a pay-as-you-go API relay

BestRelay forwards DeepSeek V4 Pro under the model name you write

Start with the part of BestRelay that makes the rest of the price card easy to read: the model you name is the model that answers. BestRelay is an independent API relay. It takes each request and forwards it to the model provider under the model name the caller wrote, so nothing is quietly swapped for a cheaper tier on the way through. If you write DeepSeek V4 Pro, DeepSeek V4 Pro is what runs. That sounds like a small promise, but on a relay it is the one that decides whether a price card means anything.

The line-up is curated: 20 public models, each listed under its own name with four prices — input, output, cache write and cache read. The naming is deliberate, because a price without a name attached is just a number. When a provider retires a model id, BestRelay delists it, so the list stays a record of what can actually be called. For a reader planning a day's work, that matters in a plain way: you choose the model by its published name, you read its rates, and you can look afterwards in the usage records to see which model answered and what it cost.

That is also why this article reads one line at a time rather than describing BestRelay in adjectives. A curated line-up of 20 public models, each with its own rates and its own name, is a small thing to hold in your head. The rest of this piece takes DeepSeek V4 Pro as the working example and follows it through the card, the balance, the clients and the records.

The DeepSeek V4 Pro line on the BestRelay rate card

DeepSeek V4 Pro is priced per 1,000,000 tokens: input $1.53, output $4.58, cache read $0.0514. Read in order, those three figures say how the day will be billed. Input is the text you send; output is the text the model sends back; cache read is the cheaper rate for material the provider can serve from cache rather than treating as fresh input. A cached read at $0.0514 sits well below the $1.53 input rate, so a workload that repeats a fixed body of context has a clear reason to arrange its requests around that line.

Billing on BestRelay runs through 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. Nothing is held in advance, which means the balance you see is the balance you have; there is no second figure waiting to be deducted later.

Two failure cases are worth reading carefully, because they are the ones that quietly cost money on other setups. Requests the provider rejects are recorded at zero. When the provider cannot be reached, nothing is charged. So an error is not billed as if it were a successful answer, and an unreachable provider does not drain a prepaid balance. The rate card and the billing rule agree with each other: you pay for tokens the provider reports on a call that ran.

What a day's DeepSeek V4 Pro traffic costs on BestRelay

The card is easier to judge with an example. Take a day of 500,000 input and 100,000 output tokens on DeepSeek V4 Pro. As an example, that day costs $1.22 on BestRelay. Twenty-two such days cost $26.89. If 350,000 of the input is read from cache instead, the day costs $0.71. Nothing about those figures requires a plan bought in advance; they are simply the card applied to a stated amount of traffic.

Money enters the account in plain amounts. Each new account opens with $0.35 of credit. Top-ups run through Stripe checkout at $1.39, $6.94, $13.89, $27.78, $69.44 or $138.89. The free credit alone covers about 1.6 million GPT-6.1 Sol input tokens, which is a reasonable way to see that the opening balance is small but usable.

If you top up and change your mind, 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, so treat those two as separate from ordinary balance.

Which clients can talk to DeepSeek V4 Pro through BestRelay

DeepSeek V4 Pro pairs with the wider tool set: Codex, Cursor, Cherry Studio, Cline, OpenCode, any SDK and cURL. Those tools pair only with GPT, DeepSeek, Kimi, GLM, Qwen and Grok. DeepSeek V4 Pro sits inside that group, so you can bring it to the client you already keep open rather than changing your habits to fit the model.

The Claude models are narrower: they are served only to the official Claude Code and Claude Desktop clients, version 2.1.285 or newer, and they do not work in Cursor, Cherry Studio, Cline, OpenCode, any SDK or cURL. The Gemini models use only their native protocol, through Gemini CLI or direct generateContent calls, and are not OpenAI-compatible. These are separate lanes, and the card prices each lane's models on its own terms.

Connecting a client takes two settings: the address and a key created under API keys in the dashboard. The details are in the docs. Keep the address and every setting to yourself; a key is a credential, and BestRelay lets you void one at any time. An account can hold several keys, which is convenient when you want one for a daily client and another for a script.

BestRelay records every call and keeps request text for 30 days

Every call appears in the usage records with its model, tokens and charge. That record is the practical answer to the question a rate card raises: what did this actually cost, and which model produced it? Because the model named is the model that answers, the model field in the record is worth reading rather than assuming.

Limits are stated plainly: 120 requests a minute on one key, and 8 calls in flight on one account. On privacy, the user-written text of requests is kept for 30 days to review abuse and settle billing questions; system prompts and model replies are not stored. So the record holds your prompt for a defined window, and not the model's answer.

Referrals are handled in the open as well. 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.

Taken together, BestRelay's shape is consistent: a curated line-up of 20 public models, each named and priced on its own line, a prepaid pay-as-you-go balance with no monthly fee, and records that tell you which model answered and what it cost. Start with DeepSeek V4 Pro at $1.53 input and $4.58 output per 1,000,000 tokens, and try BestRelay with the traffic you already have.