← Volver a las guías

59API FAQ: GPT-5.6 Sol monthly cards on one API relay

Do I actually need a monthly card to use 59API?

Short answer: no. A monthly card is optional, not a subscription requirement. 59API is an independent third-party AI API relay where you get one account, one API key and one API host at https://api.59api.com, then route requests to Claude, GPT, Gemini, DeepSeek and other models through it.

The default way to pay is a prepaid balance. You top up, and each request is charged per request by real token usage. There is no monthly fee, and your balance never expires. That matters if you are the kind of developer who ships in bursts: a quiet week does not cost you anything, and the money you already loaded stays there until you use it.

• Top-up amounts: ¥10, ¥50, ¥100, ¥200, ¥500, ¥1000.
• Payment methods: WeChat Pay or bank card. Invoices are available.
• Billing is in CNY; USD figures are shown at ¥7.2 = $1.
• All model prices are per 1 million tokens.

Monthly and weekly cards sit alongside this model rather than replacing it. If your usage is steady and predictable, a card can give you a defined quota and a defined validity window. If your usage is unpredictable, the prepaid balance is the simpler choice. Either way, you are working with the same 59API account and the same host, so you can start with a small top-up and decide later whether a card makes sense.

What does a monthly card actually give me?

A monthly card is a separate key with its own rules, and those rules are worth reading carefully before you buy one. Here is the short version of how card mechanics work on 59API.

• A card has its own key and can only call that family of models.
• A card is paid only from the card. It never falls back to wallet balance.
• A balance key never draws on a card.
• Unused quota does not roll over.
• Quota and caps are card metering units written in $, not official list-price dollars, so they should not be converted into official prices.
• Usage drains at the card metering price × 2.4.

That last point is the one developers ask about most often. The quota figure on the card is a metering quantity, not a promise of official-list-price output, and the reference values shown alongside each card are informational comparisons rather than a separate balance you can spend. Weekly cards use the same metering logic, just over 7 days instead of 30.

So the practical question is simple: will you use one model family heavily enough to fill a fixed quota inside its window? If yes, a card gives you a predictable ceiling. If your workload rotates across model families, or spikes and fades, the prepaid balance avoids the rollover question entirely because it never expires.

Which monthly cards are worth looking at?

There are several tiers, and the differences are quota size, weekly cap and price. Rather than list every option, here are representative monthly cards to compare.

• C-PRO 入门版 (Claude, 30 days): $83.19 (¥599); quota $610, weekly cap $230; reference value at official list prices about $169.44 (¥1220).
• C-MAX 标准版 (Claude, 30 days): $319.31 (¥2299); quota $2350; reference value at official list prices about $652.78 (¥4700).
• G-PRO 入门版 (GPT, 30 days): $83.19 (¥599); quota $2300, weekly cap $880; reference value at official list prices about $638.89 (¥4600).
• G-PRO 标准版 (GPT, 30 days): $160.42 (¥1155); quota $4500, weekly cap $1700; reference value at official list prices about $1250 (¥9000).
• G-PRO 企业标准版 (GPT, 30 days): $555.42 (¥3999); quota $15550; reference value at official list prices about $4319.44 (¥31100).

All of these drain at the card metering price × 2.4, and all unused quota is lost rather than rolled over. If you want a shorter window instead, the 【周卡】C-PRO $290 (Claude, 7 days) card is $40.14 (¥289) with quota $290, and the 【周卡】G-PRO $1100 (GPT, 7 days) card is $40.56 (¥292) with quota $1100.

Pick the tier by expected volume, not by headline quota. A card you drain in four days is money you spent early; a card you never fill is money you spent late. Both are avoidable with a plain prepaid balance.

How do cards compare with pay-as-you-go on 59API?

Pay-as-you-go on 59API is billed by real token usage at rates that sit below official list prices for the models where a list price exists. For example, GPT-5.6 Sol is $0.3472 (¥2.5) per 1 million input tokens, $1.94 (¥14) per 1 million output tokens and $0.0278 (¥0.2) for cache read. GPT-6.1 Sol is $0.1847 (¥1.33) input and $0.9264 (¥6.67) output per 1 million tokens.

Two worked examples make the comparison concrete. For GPT-5.6 Sol, 2M input plus 0.5M output tokens without cache costs $1.67 (¥12) on 59API, versus official list price $18 (about ¥129.6), so 59API is about 9% of list price. For GPT-6.1 Sol, 5M input plus 1M output tokens without cache is $1.85 (¥13.32) on 59API, against official list price $20 (about ¥144), so again about 9% of list price.

That is the trade-off in one line: cards give you a fixed quota and a fixed window, while pay-as-you-go gives you flexibility and no expiry. If your usage in a month is roughly predictable, a card can simplify budgeting. If it is not, per-token billing keeps the cost tied to what you actually send. Official list prices here are reference prices only, and 59API is an independent third-party relay, not an official channel.

How do I connect and start using a card?

Connecting is where most setup friction disappears, because 59API uses one host and one key format. For Claude models, use the genuine Claude Code or Claude Desktop client, version 2.1.285 or newer, and set ANTHROPIC_BASE_URL=https://api.59api.com and ANTHROPIC_AUTH_TOKEN=YOUR_59API_KEY. Claude models work only in that client, never through the OpenAI SDK, cURL or other clients.

For GPT, DeepSeek and other non-Claude, non-Gemini models, use base_url https://api.59api.com/v1 with the chat endpoint https://api.59api.com/v1/chat/completions. That path covers Codex, OpenAI SDKs, cURL and OpenAI-compatible clients such as Cursor, Cherry Studio, Cline and OpenCode. For Gemini, use GOOGLE_GEMINI_BASE_URL=https://api.59api.com and GEMINI_API_KEY=YOUR_59API_KEY in Gemini CLI, because Gemini models use the native protocol and do not work on /v1/chat/completions.

Whether you paste a balance key or a card key, the workflow is the same: pick the key that matches the model family you are calling, point the client at the right base URL, and start sending requests. Start with a small top-up if you are unsure about volume, and treat a monthly card as the upgrade for usage you can already predict.

Docs