GPT-5.6 Sol API relay pricing at Relay365: what $0.25 buys

What the $0.25 signup credit buys on the GPT-5.6 Sol lane

Relay365 operates as a round-the-clock duty desk for AI APIs, and every new account receives a signup credit of $0.25 to try a lane before topping up. The first step is to create the account, log in to the dashboard, and generate an API key with a tag so you can tell this test key apart from any others. Then bind the key to the lane you want and copy the quick-wiring snippet the dashboard provides, which points at the Relay365 base URL. On the GPT-5.6 Sol lane (wire name gpt-5.6-sol), that small credit covers a surprising amount of work because Relay365 prices the model far below the official list. The rates at Relay365 are $0.3736 per 1 million input tokens, $2.09 per 1 million output tokens, and $0.0292 per 1 million cache-read tokens. The official list price for GPT-5.6 Sol is $4 input, $20 output, and $0.4 cache read, which means the Relay365 output rate is about 10% of official.

Relay365 is an independent third-party relay, so the account is a prepaid wallet kept in CNY with approximate USD rates, there is no monthly fee, and every call is settled on the token counts the upstream reports. Because billing is per call and per token type, you can predict what the credit buys before you spend it. A call that fails upstream with a 502 or 503 is not billed, and a disconnected stream is billed only for what was actually used, so the credit is not wasted on failed attempts. Within 14 days of a top-up, unused balance can be refunded minus what was consumed and the payment channel's fees, though the signup credit itself is promotional and not refundable. If you hit a limit, remember the HTTP 429 rule: more than 120 calls in 60 seconds on one key, or more than 8 requests in flight for one account, will be rejected, so pace your first tests.

A worked bill, call by call on GPT-5.6 Sol

For example, if you send 500,000 input tokens and receive 30,000 output tokens in a day, the Relay365 bill works out line by line like this. The input line is 500,000 divided by 1,000,000, times $0.3736, which is $0.1868. The output line is 30,000 divided by 1,000,000, times $2.09, which is $0.0627. Added together, the day costs $0.2495, so the $0.25 signup credit covers the whole day with a fraction left over. Check the arithmetic yourself in the usage log: the log shows the input, output, cache recall, and charge for each of the latest 200 requests, and the charge column should match these sums exactly.

For example, if your workload leans on cache reads, 8,000,000 cache-read tokens at $0.0292 per 1 million tokens cost 8 times $0.0292, which is $0.2336, again inside the credit. For example, if you send 200,000 input tokens, receive 20,000 output tokens, and recall 1,000,000 cache-read tokens, the lines are $0.07472 for input, $0.0418 for output, and $0.0292 for cache read, totaling $0.14572. For example, spread across three days at 150,000 input and 10,000 output tokens each day, each day costs $0.05604 for input plus $0.0209 for output, or $0.07694, and three days total $0.23082, still within the $0.25 credit. These are examples only, but they show how the Relay365 wallet stretches the signup credit across real calls on the GPT-5.6 Sol lane, and each example can be verified against the per-call charge in the log.

Four token types priced separately on one table

Every lane at Relay365 sits on one table, and the four token types are priced separately: input, output, cache write, and cache read. GPT-5.6 Sol has no cache-write rate, but it does have a cache-read rate of $0.0292 per 1 million tokens, so a long conversation that recalls cached context is billed on that separate line. Other lanes price all four; for example, Claude Sonnet 5.5 (wire name claude-sonnet-5-5) is $1.18 input, $5.91 output, $1.48 cache write, and $0.0944 cache read per 1 million tokens, with the Relay365 output rate about 59% of the official list price. Because Relay365 is an independent third-party relay, you compare these numbers against the official list prices for Claude and GPT lanes directly on the pricing page, and the full table shows every lane side by side, so you can pick the lane whose token mix fits your workload before you spend the credit.

To check a bill on any lane, multiply each token count by its rate per 1 million and add the lines. For example, on Claude Sonnet 5.5, 1,000,000 input tokens cost $1.18, 200,000 output tokens cost $1.182, 500,000 cache-write tokens cost $0.74, and 2,000,000 cache-read tokens cost $0.1888, for a total of $3.2908 on Relay365. The same call at the official list price would be $2 input, $2 output, $2 cache write, and $0.2 cache read, or $6.20, so the Relay365 output rate being about 59% of official is visible in the comparison. Relay365 settles every call on the token counts the upstream reports, so the four lines in the usage log reconcile exactly with these sums.

Tracing every request in the Relay365 usage log

The Relay365 dashboard keeps a usage log of the latest 200 requests, showing the time, model, input, output, cache recall, charge, and result of each call. That means every request can be traced from the duty desk: open the log, find the row for your test call, and read off the token counts and the charge, then confirm the charge equals the sum of the priced lines. The log does not show which key made the call, so use key tags to keep your experiments organized. Alongside the log, the ledger shows the wallet, bounty, and pass accounts side by side, so the $0.25 signup credit sits in its own place and you can watch it draw down call by call. Relay365 settles every call on the token counts the upstream reports, so the charge you see in the log matches the line-by-line sums you worked out. The dashboard also offers API keys with tags, a key bound to a lane, quick-wiring snippets, in-site messages to the desk (log in first), and six interface languages.

When a duty pass fits instead of the wallet

For steady workloads, Relay365 also offers optional duty passes for Claude and GPT lanes only, which come with a dedicated key that never touches the wallet, so pass usage and wallet usage stay cleanly separated in the ledger. For example, the GPT 30-day pass with $3000 of quota and a $900 weekly cap costs $81.53 for 30 days, and draws up to $3000 of quota at the duty-pass rate times 2.5, at most about $1200 of usage at list prices, with at most $900 of quota per 7 days. For example, the Claude 14-day pass with $700 of quota and a $380 weekly cap costs $50.97 for 14 days, drawing up to $700 of quota at the duty-pass rate times 2.5, at most about $280 of usage at list prices, with at most $380 of quota per 7 days. Other models always use the wallet, so the pass only matters once your GPT-5.6 Sol or Claude volume is steady enough that a fixed monthly cost beats pay-as-you-go. These passes are optional; the prepaid wallet with the $0.25 signup credit is enough to start, run the worked bill yourself, and decide later whether a pass fits.

Start at the Relay365 duty desk

Relay365 keeps the desk open around the clock, with one table for every lane, a traceable log for every request, and separate pricing for all four token types, so a new account can go from signup to a verified per-call bill in minutes. Create a Relay365 account today, take the $0.25 signup credit, send your first GPT-5.6 Sol calls, and watch the bill add up line by line in the usage log.