开发者文档

本页只覆盖对客户公开的开发者接口：推理 API、购卡/充值/查单、自助额度查询和 Key 管理。后台管理路由、支付 webhook、运维排查接口不在公开文档范围内。

购买决策前先看这 4 点

主调用方式是 OpenAI 兼容的 POST /v1/chat/completions。
支持 Python、Node.js、curl 直接接入。
首购成功后自动交付首个 API Key；已有 API Key 可直接充值。
平台提供购卡、充值、查单、额度查询、日志查询和 Key 管理接口。

购买 API 已有 API Key？去充值

接入地址

站点域名

https://llm-agent.cc客户端只接受域名、站点地址或 API Host 时使用

OpenAI Base URL

https://llm-agent.cc/v1OpenAI SDK 和支持 Base URL 的客户端优先使用

自定义端点 URL

https://llm-agent.cc/v1/chat/completions客户端只能填写完整端点时使用

鉴权方式

Authorization: Bearer sk-your-api-keyAPI Key 以 sk- 开头，购买成功后自动交付

某些客户端在 URL 校验阶段就会报错。这种情况下不要先怀疑 Key，先检查地址类型是否填对：支持 Base URL 的客户端填 https://llm-agent.cc/v1；只能填完整端点的客户端填 https://llm-agent.cc/v1/chat/completions；只接受站点地址的客户端填 https://llm-agent.cc。

推理 API 参考

POST/v1/chat/completions

摘要

OpenAI-compatible chat completions endpoint used by third-party SDKs and clients.

请求格式

POST /v1/chat/completions
Content-Type: application/json
Authorization: Bearer sk-your-api-key

{
  "model": "your-model-id",
  "messages": [
    { "role": "system", "content": "You are a concise assistant." },
    { "role": "user", "content": "Give me a one-line introduction." }
  ],
  "temperature": 0.7
}

字段	类型	必填	说明
`model`	string	Yes	Upstream model identifier exposed by your runtime.
`messages`	array	Yes	Chat messages in OpenAI format.
`temperature`	number	No	Sampling temperature. Common range: 0 to 2.
`max_tokens`	integer	No	Upper bound for generated tokens.
`stream`	boolean	No	Enable streaming responses when supported upstream.
`user`	string	No	Optional caller identifier for tracing and abuse controls.

成功响应格式

{
  "id": "chatcmpl_01JEXAMPLE",
  "object": "chat.completion",
  "created": 1712380800,
  "model": "your-model-id",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "This endpoint is compatible with the OpenAI chat format."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 21,
    "completion_tokens": 16,
    "total_tokens": 37
  }
}

字段	类型	说明
`id`	string	Unique completion identifier.
`choices`	array	Returned candidate completions.
`choices[].message`	object	Assistant reply content.
`usage`	object	Prompt, completion, and total token counts.
`model`	string	Model id used for the request.

多语言 SDK 示例

先用下面这些最小示例跑通，再迁移到自己的业务代码或第三方客户端。

支持 7+ 种编程语言：Python、Node.js、Go、Java、PHP、C# 和 cURL。

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-api-key",
    base_url="https://llm-agent.cc/v1"
)

response = client.chat.completions.create(
    model="your-model-id",
    messages=[
        {"role": "system", "content": "You are a concise assistant."},
        {"role": "user", "content": "Give me a one-line introduction."}
    ]
)

print(response.choices[0].message.content)

流式输出（实时响应）示例

启用流式输出可以实时接收生成的 token，显著提升用户体验。

为什么要使用流式输出？

降低感知延迟 - 用户无需等待完整响应，可以立即看到内容
更好的长文本体验 - 内容像人类打字一样逐步显示
成本相同 - 与同步输出成本完全一致，只是传输方式不同

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-api-key",
    base_url="https://llm-agent.cc/v1"
)

stream = client.chat.completions.create(
    model="your-model-id",
    messages=[
        {"role": "system", "content": "You are a concise assistant."},
        {"role": "user", "content": "Give me a one-line introduction."}
    ],
    stream=True  # Enable streaming
)

# Print tokens as they arrive
for chunk in stream:
    if chunk.choices[0].delta.content is not None:
        print(chunk.choices[0].delta.content, end="")
print()  # Newline at the end

平台 REST API 参考

下面是对客户公开的站内辅助接口，用于购买、充值、查单、额度查询和 Key 管理。

GET/api/topup/packages

鉴权: Public

List fixed recharge packages currently available for existing API keys.

请求示例

GET /api/topup/packages

响应示例

{
  "ok": true,
  "data": [
    {
      "id": "pkg_100",
      "code": "topup_100",
      "name": "100 CNY package",
      "currency": "CNY",
      "priceCents": 10000,
      "baseQuotaAmount": 6000000000,
      "bonusQuotaAmount": 1000000000,
      "quotaAmount": 7000000000
    }
  ]
}

说明

Use this endpoint to render public recharge options without hardcoding them in the client.

POST/api/purchase/orders

鉴权: Public

Create a first-purchase order for a plan code and reserve delivery contact details.

请求示例

{
  "email": "[email protected]",
  "planCode": "starter",
  "sourcePage": "pricing"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "BUY202604060930ABCD",
    "amountCents": 29900,
    "contactEmail": "[email protected]",
    "payUrl": "/pricing/pay?orderNo=BUY202604060930ABCD&contactEmail=buyer%40example.com"
  }
}

说明

Send x-idempotency-key when retrying create-order requests.
The response shape is a trimmed example; extra payment and delivery metadata may be included.

POST/api/purchase/orders/pay

鉴权: Public

Prepare a payment session for an existing first-purchase order.

请求示例

{
  "orderNo": "BUY202604060930ABCD",
  "contactEmail": "[email protected]"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "BUY202604060930ABCD",
    "provider": "xunhupay",
    "checkoutUrl": "https://pay.example.com/checkout/abc",
    "qrcodeUrl": "https://pay.example.com/qrcode/abc"
  }
}

说明

Use the order number and contact email pair returned from order creation.

POST/api/purchase/orders/query

鉴权: Public

Query a first-purchase order by order number and contact email.

请求示例

{
  "orderNo": "BUY202604060930ABCD",
  "contactEmail": "[email protected]"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "BUY202604060930ABCD",
    "status": "PAID",
    "deliveryState": "DELIVERED",
    "contactEmail": "[email protected]"
  }
}

说明

Returns 404 with purchase_order_not_found when the pair does not match an order.

POST/api/topup/preview

鉴权: Public

Verify that an API key exists and can receive additional credits.

请求示例

{
  "apiKey": "sk-live_example"
}

响应示例

{
  "ok": true,
  "data": {
    "tokenName": "Production Key",
    "availableQuota": 5400000000,
    "usedQuota": 1600000000,
    "totalQuota": 7000000000
  }
}

说明

Clients typically call this before showing recharge package selection.

POST/api/topup/orders

鉴权: Public

Create a recharge order for an existing API key.

请求示例

{
  "apiKey": "sk-live_example",
  "packageCode": "topup_100",
  "contactEmail": "[email protected]",
  "couponCode": "SPRING10"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "TOP202604060935WXYZ",
    "amountCents": 9000,
    "contactEmail": "[email protected]",
    "payUrl": "/topup/pay?orderNo=TOP202604060935WXYZ&contactEmail=buyer%40example.com"
  }
}

说明

Either packageCode or tokenUnits is expected depending on the current pricing mode.

POST/api/topup/orders/pay

鉴权: Public

Prepare a payment session for an existing recharge order.

请求示例

{
  "orderNo": "TOP202604060935WXYZ",
  "contactEmail": "[email protected]"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "TOP202604060935WXYZ",
    "provider": "xunhupay",
    "checkoutUrl": "https://pay.example.com/checkout/topup"
  }
}

说明

The payment provider payload may include QR code and mobile URL variants.

POST/api/topup/orders/query

鉴权: Public

Query a recharge order by order number and contact email.

请求示例

{
  "orderNo": "TOP202604060935WXYZ",
  "contactEmail": "[email protected]"
}

响应示例

{
  "ok": true,
  "data": {
    "orderNo": "TOP202604060935WXYZ",
    "status": "PAID",
    "credited": true
  }
}

说明

Returns 404 with topup_order_not_found when the pair does not match an order.

POST/api/quota/summary

鉴权: Public

Return current credit totals and billing-derived balance information for an API key.

请求示例

{
  "apiKey": "sk-live_example"
}

响应示例

{
  "ok": true,
  "data": {
    "tokenName": "Production Key",
    "totalTokens": 7000000000,
    "usedTokens": 1600000000,
    "remainingTokens": 5400000000,
    "responseTimeMs": 283
  }
}

说明

Returns INVALID_KEY, RATE_LIMIT, SERVER_ERROR, NETWORK, or UNKNOWN in the error payload.

POST/api/quota/logs

鉴权: Public

Return usage log entries for an API key, including cache hit metadata and response times.

请求示例

{
  "apiKey": "sk-live_example"
}

响应示例

{
  "ok": true,
  "data": {
    "items": [
      {
        "id": "log_01",
        "modelName": "gpt-4o-mini",
        "quota": 3200,
        "hasCacheHit": false
      }
    ],
    "total": 1,
    "truncated": false
  }
}

说明

The backend can cap or truncate very large log histories.

GET/api/tokens

鉴权: Session cookie required

List API keys owned by the currently signed-in dashboard user.

请求示例

GET /api/tokens

响应示例

{
  "ok": true,
  "items": [
    {
      "id": "key_01",
      "displayName": "Default Key",
      "maskedKey": "sk-****abcd",
      "status": "ACTIVE"
    }
  ]
}

说明

Used by the dashboard key-management page after sign-in.

POST/api/tokens

鉴权: Session cookie required

Create a new API key for the currently signed-in dashboard user.

请求示例

{
  "displayName": "Default Key"
}

响应示例

{
  "ok": true,
  "token": {
    "id": "key_01",
    "displayName": "Default Key",
    "maskedKey": "sk-****abcd",
    "status": "ACTIVE"
  },
  "plainKey": "sk-live_once_only"
}

说明

The full plainKey is only returned once at creation time.

PATCH/api/tokens/{id}/disable

鉴权: Session cookie required

Disable an API key owned by the currently signed-in dashboard user.

请求示例

PATCH /api/tokens/key_01/disable

响应示例

{
  "ok": true,
  "token": {
    "id": "key_01",
    "status": "REVOKED"
  }
}

说明

Returns 404 when the key id does not belong to the current user.

DELETE/api/tokens/{id}

鉴权: Session cookie required

Delete an API key owned by the currently signed-in dashboard user.

请求示例

DELETE /api/tokens/key_01

响应示例

{
  "ok": true
}

说明

Deletion is irreversible and updates the linked upstream token state.

错误码对照表

推理 API

HTTP	名称	含义	处理建议
`400`	INVALID_REQUEST	Malformed JSON, invalid parameters, or unsupported request shape.	Validate the body and compare it with the OpenAI-compatible schema.
`401`	UNAUTHORIZED	Missing, expired, or invalid API key.	Check the Bearer token and verify the key still exists.
`429`	RATE_LIMIT	Too many requests were sent in a short time window.	Retry with backoff and reduce burst traffic.
`500`	UPSTREAM_ERROR	The upstream runtime or proxy returned an internal error.	Retry later and inspect status/monitoring if the issue persists.

平台 REST API

错误码	HTTP	含义
`invalid_payload`	400	The JSON body did not match the endpoint schema.
`purchase_order_create_failed`	400	The first-purchase order could not be created.
`purchase_pay_failed`	400	Payment session creation for a first purchase failed.
`purchase_order_not_found`	404	No matching purchase order was found for the order number/email pair.
`topup_preview_failed`	400	The target API key could not be verified for top-up.
`topup_order_create_failed`	400	The recharge order could not be created.
`topup_pay_failed`	400	Payment session creation for a recharge failed.
`topup_order_not_found`	404	No matching top-up order was found for the order number/email pair.
`INVALID_KEY`	400	The supplied API key was not accepted by the quota query backend.
`RATE_LIMIT`	500	The quota query upstream is currently throttling requests.
`SERVER_ERROR`	500	The quota query upstream is temporarily unavailable.

限流说明

Authentication, password reset, and quota lookup endpoints include cooldown and lightweight anti-abuse protection.
Treat 429 responses as retryable and apply exponential backoff on the client side.
For checkout APIs, reuse the same x-idempotency-key when retrying order creation after a network failure.
Burst traffic and scripted retries should be throttled client-side to avoid temporary lockouts.

模型与定价

模型可用性

Model availability is managed by the upstream newAPI / CLI proxy runtime, so the exact model list can change independently of this website.
Use the model id exposed by your upstream runtime when calling /v1/chat/completions.
The pricing table below describes purchase and top-up packages sold on this site, not per-model token billing.

当前套餐定价

类型	套餐	价格	原额度	赠送	到账总额	说明
Purchase	20 元首购套餐	¥20.00	1B	0B	1B	First-time package that provisions the initial API key after payment.
Purchase	40 元首购套餐	¥40.00	2B	0B	2B	First-time package that provisions the initial API key after payment.
Purchase	60 元首购套餐	¥60.00	3B	0B	3B	First-time package that provisions the initial API key after payment.
Purchase	100 元首购套餐	¥100.00	5B	1B	6B	First-time package that provisions the initial API key after payment.
Purchase	200 元首购套餐	¥200.00	10B	2B	12B	First-time package that provisions the initial API key after payment.
Purchase	1000 元首购套餐	¥1,000.00	50B	10B	60B	First-time package that provisions the initial API key after payment.
Top up	20 元充值套餐	¥20.00	1B	0B	1B	Recharge package for an existing API key.
Top up	40 元充值套餐	¥40.00	2B	0B	2B	Recharge package for an existing API key.
Top up	60 元充值套餐	¥60.00	3B	0B	3B	Recharge package for an existing API key.
Top up	100 元充值套餐	¥100.00	5B	1B	6B	Recharge package for an existing API key.
Top up	200 元充值套餐	¥200.00	10B	2B	12B	Recharge package for an existing API key.
Top up	1000 元充值套餐	¥1,000.00	50B	10B	60B	Recharge package for an existing API key.

上表来自当前仓库的套餐配置。若后续套餐、赠送策略或货币发生调整，应同步更新文档和购买页。

加载中...