Claude Code pricing on one table: the Relay365 API relay walkthrough
Opening the Relay365 dashboard: keys before models
When you open a Relay365 account, the duty desk starts you on the pages you can actually use. The first is the key page: create an API key, give it a tag so it is easy to recognise later, and bind it to the lane you plan to call. That binding is the decision worth making first, because it decides which models the key can reach and keeps the rest of the table out of your way. Beside it, the dashboard offers quick-wiring snippets, so you can copy a working request shape into your own project instead of assembling headers and endpoints from scratch.
Relay365 is an independent third-party relay, run like a round-the-clock duty desk, and the interface is available in six languages. A new account also starts with $0.25 in signup credit, small but immediately usable for a test request.
Then you arrive at the page this walkthrough is really about: the line table. Every lane sits on one page, and that table is the map you keep open while you build.
Reading the line table: four token types priced apart
The table lists each model with its wire name and four separate prices, because input, output, cache write and cache read are four different kinds of work. A long prompt is not priced like a long answer, and a cached read is not priced like a fresh write. Keeping them apart is what lets you estimate a request before you send it.
Claude Fable 5 (claude-fable-5) shows how the table reads: input $3.79, output $18.94, cache write $4.66 and cache read $0.2917 per 1 million tokens. Against the official list price of $10 input, $50 output and $1 cache read, the Relay365 output rate is about 38% of official. GPT-6 Astra (gpt-6-astra) is listed at input $0.7444, output $3.72 and cache read $0.075, and its output rate is about 7% of the official list price of $50.
For families where no official price is quoted, the table simply shows the Relay365 rates. Gemini 3.1 Pro (gemini-3.1-pro) is input $1.54, output $8.94, cache write $1.54 and cache read $0.1542, while DeepSeek V4 Pro (deepseek-v4-pro) is input $1.45, output $4.35 and cache read $0.0486. The pricing page carries every lane and every rate, so you can check the numbers yourself before you choose.
A quick example, using only those rates: if you send 100,000 input tokens and 50,000 output tokens a day to Claude Fable 5, that is $0.379 of input and $0.947 of output, or about $1.33 a day. The same daily shape on GPT-6 Astra works out at about $0.26, and on Gemini 3.1 Pro, 200,000 input tokens plus 100,000 output tokens comes to about $1.20 a day. The account itself is kept in CNY, while these rates are shown in approximate USD per 1 million tokens.
Matching a lane to the client you actually use
Once a lane is chosen, the client matters, and the rules at Relay365 are firm. Claude models work only from the official Claude Code and Claude Desktop clients, with Claude Code 2.1.285 or newer. Cursor, Cherry Studio, Cline, OpenCode, any SDK and cURL pair only with GPT, DeepSeek, Kimi, GLM, Qwen, Grok, MiniMax and Hunyuan lanes. Gemini runs on its native protocol only, through Gemini CLI: :generateContent for a single reply and :streamGenerateContent?alt=sse for a stream, and it cannot go through /v1/chat/completions.
The desk also watches call volume on your behalf. More than 120 calls in 60 seconds on one key, or more than 8 requests in flight for one account, will return HTTP 429. The quick-wiring snippets on the key page show the request shape each lane expects, which is the fastest way to avoid a mismatch between client and protocol before it costs you a call.
The prepaid wallet and the duty passes beside it
Relay365 bills from a prepaid wallet, with no monthly fee. Every call is settled on the token counts the upstream reports. A call that fails upstream with a 502 or 503 is not billed at all, and a disconnected stream is billed only for what was actually used.
Duty passes sit alongside the wallet as an option for Claude and GPT lanes only; other models always use the wallet. A pass comes with a dedicated key that never touches the wallet. The Claude 14-day pass costs $50.97 for 14 days and draws up to $390 of quota at the duty-pass rate x2.5, which is at most about $156 of usage at list prices and less on models whose pass rate sits above list, with at most $210 of quota per 7 days. The GPT 30-day pass costs $81.53 for 30 days and draws up to $2350 of quota at the duty-pass rate x2.5, at most about $940 of usage at list prices and less on models whose pass rate sits above list, with at most $700 of quota per 7 days.
A call already let through on a pass is drawn in full from the pass even if it runs past the quota or the 7-day ceiling, and the overrun counts against the earliest-expiring card. When a pass is nearly used up, each running call holds $3 x the draw rate, and an extra parallel call gets HTTP 429 until the running ones finish.
Top-ups are straightforward too. Within 14 days of a top-up, unused balance can be refunded minus what was consumed and the payment channel's fees; bounty and promotional credit are not refundable. The referral bounty pays 18% at level 1 and 6% at level 2 of what referred members pay, and it can be withdrawn from $11.11 or moved into the wallet.
The usage log, the ledger and the desk behind them
After the calls start, the last page worth opening is the usage log. It shows the latest 200 requests, and each row carries the time, model, input, output, cache recall, charge and result of the call. It does not show which key made the call, so treat it as a request-level record rather than a key-level one. For a relay that prices four token types separately, that level of detail is what makes a bill explainable.
Next to the log sits the ledger, where wallet, bounty and pass accounts are kept side by side, and in-site messages let you reach the desk directly once you are logged in. If a figure does not look right, that is the path to ask about it.
The order is simple: open the key page, pick the lane, read the table, wire the client, top up or pick a duty pass, then read the log. Relay365 keeps every lane on one table so you can stay in the flow instead of hunting between price lists. When you are ready, create a Relay365 account and start with a single request.