BestRelay pay-as-you-go pricing for Claude Sonnet 5.5 at $1.39 input

BestRelay forwards each request under the model name you write

BestRelay is an independent API relay. It 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. The line-up is curated: 20 public models, each listed under its own name with four prices covering input, output, cache write and cache read. When a provider retires a model id, BestRelay delists it. You therefore choose from a short, maintained list rather than a shifting catalogue, and the rate card in front of you is the rate card you are charged. 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. Requests the provider rejects are recorded at zero, and when the provider cannot be reached nothing is charged. Every new account opens with $0.35 of credit.

Top-ups run through Stripe checkout in fixed 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through your link, and 4% of the top-ups of the people they invited, and commission can be moved into the balance or withdrawn to Alipay from $20.83.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. An account can hold several keys, each voidable at any time, and every call appears in the usage records with its model, tokens and charge.

Claude models are served only to Claude Code and Claude Desktop

BestRelay applies a plain client rule to the Claude family: 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. The reason is practical rather than decorative. Anthropic's own clients speak Anthropic's protocol properly, and serving the Claude models to those clients keeps caching, tool calls and streaming behaving as the provider intended. A supported client needs two settings, the address and a key created under API keys in the dashboard, and the docs set out the steps; the address and any setting are not printed here. Limits are 120 requests a minute on one key, and 8 calls in flight on one account.

The savings follow from the rate card rather than from any change in behaviour. For example, a day of 500,000 input and 100,000 output tokens on Claude Sonnet 5.5 costs $1.40 on BestRelay, and 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. BestRelay lists Claude Sonnet 5.5 at input $1.39, output $7.01, cache write $1.74 and cache read $0.1222 per million tokens, and Claude Opus 5.5 at input $1.74, output $8.13, cache write $2.01 and cache read $0.0778.

BestRelay publishes four rates for every model it lists

Because BestRelay is a relay rather than a wrapper that reshapes answers, its pricing page is the whole commercial proposition. Every entry shows input, output, cache write and cache read where the provider supports them. Claude Fable 5 is input $4.17, output $20.97, cache write $5.14 and cache read $0.3472. Claude Fable 5.1 is input $4.10, output $20.42, cache write $5.00 and cache read $0.3472. Claude Haiku 4.5 is input $0.4167, output $2.01, cache write $0.4861 and cache read $0.0333. Claude Opus 5 is input $2.29, output $11.46, cache write $2.85 and cache read $0.1389. Claude Sonnet 5 is input $1.32, output $6.67, cache write $1.67 and cache read $0.1167. On the GPT side, GPT-5.6 Sol is input $0.4167, output $2.29 and cache read $0.0333; GPT-6 Astra is input $0.7639, output $3.96 and cache read $0.0792; GPT-6.1 Sol is input $0.2083, output $1.11 and cache read $0.0431.

The rest of the line-up is equally plain. DeepSeek V4 Pro is input $1.53, output $4.58 and cache read $0.0514. Kimi K3 is input $1.39, output $7.01 and cache read $0.1389. GLM-5.3 is input $0.625, output $1.94 and cache read $0.1389. Qwen3.8 Max is input $0.8333, output $2.57 and cache read $0.1014. Grok 4.6 is input $0.625, output $1.81 and cache read $0.1389, and Grok 4.7 is input $0.625, output $1.94 and cache read $0.1389. Gemini 2.5 Flash is input $0.2778, output $2.08, cache write $0.2778 and cache read $0.0264; Gemini 3.1 Pro is input $1.74, output $9.93, cache write $1.74 and cache read $0.1389; Gemini 3.8 Flash is input $1.25, output $5.56, cache write $1.25 and cache read $0.1236; Gemini 4 Argon is input $1.67, output $8.06, cache write $1.67 and cache read $0.0847. For example, a day of 500,000 input and 100,000 output tokens on GPT-6.1 Sol 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, against $2.00, $44.00 and $1.48 at the official list price, 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.

Connecting a supported client takes two settings and a key

BestRelay keeps the integration deliberately small. A client needs two settings: the address and a key created under API keys in the dashboard. Details are in the docs, and neither the address nor any setting is printed in this article. An account can hold several keys, which is convenient when you want a separate key for a separate project, and each can be voided at any time if it stops being useful or is suspected of leaking. Every call appears in the usage records with its model, tokens and charge, so you can audit spending line by line rather than by guesswork. The limits are 120 requests a minute on one key, and 8 calls in flight on one account, which keeps a single account from being pushed beyond what the providers will comfortably carry.

Privacy is described in narrow terms. 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. BestRelay does not rewrite your prompt, add its own system message or paraphrase the answer. It forwards the request and returns the provider's reply. Which models you may use depends on the client: with Claude Code or Claude Desktop you get the Claude models, with Gemini CLI you get the Gemini models, and tools such as 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. This arrangement means each model is reached the way its provider intends, and the line-up stays short enough to read in full.

BestRelay charges only for calls that succeed

The billing model is straightforward. You open a prepaid pay-as-you-go balance, top it up through Stripe checkout, and each call that ends is charged once from the tokens the provider reports, at the published per-million rates. Nothing is held in advance, so a balance sits on the account rather than being reserved against it. There is no monthly fee, no subscription and no minimum spend. When the provider rejects a request, the call is recorded at zero. When the provider cannot be reached, nothing is charged at all. Top-ups come in six fixed amounts: $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89, so you can add a small amount for a quiet week or a larger one for a busy month without changing how the service works.

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, which is worth knowing before you move referral earnings into the balance. 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. 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. Because every call is logged with its model, tokens and charge, and because failed calls cost nothing, the record and the bill agree with each other. Try BestRelay with a named model and see how a relay that forwards exactly what you wrote behaves.