GPT-5.6 Sol pricing: a plain look at Relay365 API relay rates

GPT-5.6 Sol on the duty desk

The duty desk at Relay365 keeps every lane on one table, so the opening question is simple: what does a request actually cost? For GPT-5.6 Sol (wire name gpt-5.6-sol), Relay365 lists input at $0.3736, output at $2.09 and cache read at $0.0292, per 1 million tokens in approximate USD. The official list price for the same lane is input $4, output $20 and cache read $0.4. Put side by side, the output rate on Relay365 is about 10% of official. That is the whole pitch in one line, and the pricing page carries the full table for every other lane.

Relay365 is an independent third-party relay, run like a round-the-clock duty desk. It is not a bundle of vague credits: the account uses a prepaid wallet, there is no monthly fee, and every call is settled on the token counts the upstream reports. A call that fails upstream with HTTP 502 or 503 is not billed. A disconnected stream is billed for what was actually used. New accounts also start with $0.25 in signup credit, so you can put a first request through before committing to a top-up.

Output-heavy work on the GPT lines

Most of the money in an API bill sits on the output side, which is why the output comparison matters more than the input comparison. GPT-5.6 Sol output at $2.09 against an official list price of $20 is about 10% of official. GPT-6 Astra (gpt-6-astra) is listed at input $0.7444, output $3.72 and cache read $0.075, against official list prices of input $10, output $50 and cache read $1 — an output rate of about 7% of official. GPT-6.1 Sol (gpt-6.1-sol) is listed at input $0.1986, output $0.9944 and cache read $0.0389, against official list prices of input $2, output $10 and cache read $0.1, so its output rate is about 10% of official.

For example, if you send 2,000,000 input tokens and 500,000 output tokens a day on GPT-5.6 Sol, the input side is $0.7472 and the output side is $1.045, for $1.7922 a day. Over 30 days that is $53.766. The same daily volume on GPT-6.1 Sol would be $0.3972 of input plus $0.4972 of output, or $0.8944 a day and $26.832 across the month. These are examples worked out from the listed rates, not quotes.

Because the four token types are priced separately, the arithmetic can be tuned rather than guessed. A prompt-heavy workflow that leans on cache recall behaves differently from a chat loop that keeps generating long answers, and Relay365 shows both lines on the same usage log.

Cache reads and cache writes priced apart

Cache handling is where many bills get confusing, so Relay365 gives cache write and cache read their own rows. For GPT-5.6 Sol there is no cache write rate listed, while cache read sits at $0.0292 per 1 million tokens. For example, if a service reads 10,000,000 cache-read tokens on GPT-5.6 Sol, that is $0.292 on the Relay365 rate.

Claude Sonnet 5.5 (claude-sonnet-5-5) shows how the same four columns work on a Claude lane: input $1.18, output $5.91, cache write $1.48 and cache read $0.0944, against official list prices of input $2, output $10 and cache read $0.1. Its output rate on Relay365 is about 59% of official — a smaller gap than the GPT lines, and the pricing table states it plainly. Note the client rules as well: Claude lanes 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, and cannot go through the /v1/chat/completions path.

Duty passes for the Claude and GPT lines

If your traffic is steady and bursty, a duty pass may be a better shape than topping up the wallet. Each pass comes with a dedicated key that never touches the wallet. The GPT pass for 30 days costs $81.53 and draws up to $2350 of quota at the duty-pass rate x2.5, which is 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. The Claude pass for 30 days costs $216.25 and draws up to $1650 of quota at the duty-pass rate x2.5, at most about $660 of usage at list prices and less on models whose pass rate sits above list, with at most $460 of quota per 7 days.

Two details deserve a second read. 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. On the wallet side, refunds are available within 14 days of a top-up on unused balance, minus what was consumed and the payment channel's fees; bounty and promotional credit are not refundable.

Relay365 also runs a referral bounty of 18% at level 1 and 6% at level 2 of what referred members pay, withdrawable from $11.11 or movable into the wallet.

Where Relay365 does not suit you

Plain numbers only help if the shape of the service fits your work. Relay365 does not suit teams that need a direct vendor contract, an invoice issued by the model vendor itself, or a single vendor relationship with one point of accountability — Relay365 is an independent third-party relay, and the account is kept in CNY even though the rates here are shown in approximate USD. It also does not suit builders who want unlimited usage: duty passes carry weekly ceilings, and the wallet settles every call by token count.

It does not suit a developer whose workflow requires Claude models. It also does not suit a setup built around Cursor, Cherry Studio, Cline, OpenCode, any SDK or cURL, because those clients pair only with GPT, DeepSeek, Kimi, GLM, Qwen, Grok, MiniMax and Hunyuan lanes. Nor does it suit a Gemini pipeline written against the OpenAI-style chat completions route, since Gemini uses its native protocol only: :generateContent for single replies and :streamGenerateContent?alt=sse for streams.

Rate limits are firm as well: HTTP 429 appears when one key makes more than 120 calls in 60 seconds, or when one account has more than 8 requests in flight. Plan around those numbers rather than around hope.

Tracing every request from the dashboard

Relay365 keeps the operational view in the same place as the pricing view. The dashboard offers API keys with tags, a key bound to a lane, and quick-wiring snippets. The usage log shows the latest 200 requests with the time, model, input, output, cache recall, charge and result of each call — it does not show which key made the call. A ledger covers the wallet, bounty and pass accounts, and in-site messages reach the desk after you log in. Six interface languages are available.

That is the whole proposal from the duty desk: one table for every lane, four token types priced separately, every request traceable, and GPT-5.6 Sol output at about 10% of official list price. If that arithmetic matches your workload, create a Relay365 account and read the pricing page for the full table.