Claude Code API relay: Relay365 duty passes and quota

Handover note from the Relay365 duty desk

Colleague — before you open the console, here is the handover note from the Relay365 duty desk. Relay365 is an independent third-party relay, an AI API relay run like a round-the-clock duty desk, and the point of this note is to save you a second round of questions. Every lane sits on one table: the Claude, GPT, Gemini and DeepSeek lanes are all reachable from the same account, the same prepaid wallet and the same dashboard.

Two things to keep in your head. First, billing is prepaid. There is no monthly fee, and every call is settled on the token counts the upstream reports. A call that fails upstream with a 502 or 503 is not billed, while a disconnected stream is billed for what was actually used. Second, the four token types — input, output, cache write and cache read — are priced separately, so you can see where the money went instead of reading one blended number.

One more note on the desk: Relay365 is run as a duty desk, so every request can be traced in the usage log, and the lanes, the keys and the ledgers all live on one table.

What a duty pass actually is

Now the part you asked about. A duty pass on Relay365 is optional, and it applies to the Claude and GPT lanes only; every other model always uses the wallet. When you buy a pass, it arrives with a dedicated key, and that key never touches the wallet. The pass carries its own quota, its own weekly ceiling and its own ledger line.

The pass rate applies to the model rate with a 2.5 multiplier, and in return the pass gives you a large quota ceiling for the period. The quota is drawn at the duty-pass rate x2.5, which is at most about a set amount of usage at list prices, and less on models whose pass rate sits above list.

Two passes are worth putting in front of you now. The Claude 14-day pass costs $50.97 and carries $700 of quota with a ceiling of $380 per week. The Claude 30-day pass costs $506.53 and carries $6000 of quota with no weekly ceiling. Both are Claude passes, both come with a dedicated key, and both draw at the duty-pass rate x2.5.

Quota, weekly ceiling and how the two differ

Two numbers do the work, and it is worth keeping them apart. Quota is the total the pass can draw over its whole life; the weekly ceiling is the limit inside any 7-day window. A pass with a big quota and a weekly cap suits steady, spread-out traffic, while a pass with no weekly ceiling suits bursts.

Take the Claude 14-day pass: $700 of quota, at most $380 of quota per 7 days, $50.97 for 14 days. The quota is drawn at the duty-pass rate x2.5, which is at most about $280 of usage at list prices, and less on models whose pass rate sits above list. Because the window is 14 days, you get two 7-day slices — the ceiling applies per 7 days, not once across the whole pass.

The Claude 30-day pass at $506.53 is the other shape: $6000 of quota, no weekly ceiling, $506.53 for 30 days, drawn at the duty-pass rate x2.5, which is at most about $2400 of usage at list prices, and less on models whose pass rate sits above list. If your workload spikes and then sleeps, that is the shape that fits, because you are not cut off mid-week by a ceiling.

Keep the dedicated key on the lane it was issued for, and remember that a pass key never draws on your wallet balance: pass traffic and wallet traffic never mix on the same key.

Rates on the table and a worked example

Now the arithmetic, because a duty pass only makes sense once you know the rate underneath it. All figures are per 1 million tokens in approximate USD, and the account itself is kept in CNY.

• Claude Fable 5 (wire name claude-fable-5): input $3.79, output $18.94, cache write $4.66, cache read $0.2917.
• Claude Opus 5.5 (wire name claude-opus-5-5): input $1.61, output $7.59, cache write $1.90, cache read $0.0736.
• GPT-6 Astra (wire name gpt-6-astra): input $0.7444, output $3.72, cache read $0.075.
• Gemini 2.5 Flash (wire name gemini-2.5-flash): input $0.2389, output $1.91, cache write $0.2389, cache read $0.0236.

Where list prices are quoted for comparison, the Relay365 output rate for Claude Opus 5.5 is about 38% of the official list price, and for GPT-6 Astra it is about 7% of the official list price. The pricing page shows the full table for every lane, so you can check the figures for your own mix before you commit to a pass.

Here is the worked example. Suppose you run Claude Fable 5 traffic that sends 1 million input tokens and 0.5 million output tokens a day, with no cache write or cache read. At the Relay365 wallet rate, that is $3.79 for input plus half of $18.94, which is $9.47, so $13.26 a day. On the 14-day duty pass, the same traffic is drawn at the duty-pass rate x2.5: $33.15 a day, $464.10 across 14 days, and $232.05 in any 7-day window. That sits inside the $700 quota and under the $380 per week ceiling. This is an example of the arithmetic, not a promise about your own usage, which depends on the mix you actually send.

Client rules the desk will not bend

Before you wire anything up, the client rules are firm, and the duty desk would rather you know them now than find them at 02:00.

Claude models work only from the official Claude Code and Claude Desktop clients, Claude Code 2.1.285 or newer. Cursor, Cherry Studio, Cline, OpenCode, any SDK and cURL pair only with the GPT, DeepSeek, Kimi, GLM, Qwen, Grok, MiniMax and Hunyuan lanes. Gemini runs on its native protocol only, through Gemini CLI: :generateContent for single replies and :streamGenerateContent?alt=sse for streams, and it cannot go through /v1/chat/completions.

So if your plan is Claude Code, a duty pass with its dedicated key is a natural fit. If your plan is a Cursor or cURL workflow, the wallet is the route — and the same Relay365 dashboard still keeps every lane on one table.

Traceability, limits and the ledger

Traceability is the part the desk cares about most. The Relay365 dashboard gives you API keys with tags, a key bound to a lane, quick-wiring snippets, and a usage log of the latest 200 requests showing the time, model, input, output, cache recall, charge and result of each call. Note that the log does not show which key made the call, so keep your own naming tidy with tags. There is also a ledger for wallet, bounty and pass accounts, in-site messages to the desk once you are logged in, and six interface languages.

Two operating notes. Rate limits return HTTP 429 for more than 120 calls in 60 seconds on one key, or more than 8 requests in flight for one account, so plan your concurrency accordingly. And refunds are available within 14 days of a top-up: unused balance can be refunded minus what was consumed and the payment channel's fees, while bounty and promotional credit are not refundable.

If you bring colleagues along, the referral bounty pays 18% at level 1 and 6% at level 2 of what referred members pay, withdrawable from $11.11 or moved into the wallet.

That is the handover. Relay365 keeps the lanes, the quota, the dedicated keys and the ledger on one table, so you can see what you spent and why. When you are ready to try it, create a Relay365 account and start with the $0.25 signup credit.