The Token · 接入文档 / Access Docs

你的卡片Your card Endpoint · Key · Model

你的卡片上印有三个值：接入地址、你的 Key、你的 Model。把它们填进任意 OpenAI 兼容客户端即可。每张 Key 不同，Model 以卡片印刷为准。Your card shows three values: the Endpoint, your Key, and your Model. Put them into any OpenAI-compatible client. Every key is unique; use exactly the model printed on your card.

接入地址 Base URLBase URL

https://api.thetoken.biz/openai/v1

你的 KeyYour Key

YOUR_KEY （卡片印刷，每张不同）(printed on your card, unique)

你的 ModelYour Model

YOUR_MODEL （卡片印刷的模型名）(the model name on your card)

鉴权头Auth header

Authorization: Bearer YOUR_KEY

接口Endpoints

/chat/completions · /responses

额度响应头Quota headers

x-token-remaining · x-token-limit · x-token-used

Key 鉴权方式Key authentication

三选一，任意客户端都可用。Pick any one — all clients supported.

HTTP headers

# Option 1 — OpenAI standard (recommended)
Authorization: Bearer YOUR_KEY

# Option 2 — Azure style
api-key: YOUR_KEY

# Option 3 — APIM style
Ocp-Apim-Subscription-Key: YOUR_KEY

接入地址必须以 /openai/v1 结尾，客户端会自动追加 /chat/completions 或 /responses。The Base URL must end with /openai/v1; clients append /chat/completions or /responses automatically.

opencode CLI verified

新增一个 OpenAI 兼容 provider，把卡片上的 Key 与 Model 填进去。Add an OpenAI-compatible provider and fill in the Key and Model from your card.

① 编辑配置文件① Edit the config file

路径Path

~/.config/opencode/opencode.json
Windows: C:\Users\<you>\.config\opencode\opencode.json

② 加入 provider（Key 鉴权）② Add the provider (key auth)

opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "thetoken": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "The Token",
      "options": {
        "baseURL": "https://api.thetoken.biz/openai/v1",
        "apiKey": "YOUR_KEY"
      },
      "models": {
        "YOUR_MODEL": { "name": "YOUR_MODEL", "tool_call": true }
      }
    }
  }
}

③ 运行③ Run

shell

opencode run --model thetoken/YOUR_MODEL "What is 2+2?"

⚠ agent 模式每轮都会重发系统提示与工具定义，较耗 token；保存后需重启 opencode。⚠ Agent mode resends the system prompt + tool schemas every turn (token-heavy); restart opencode after saving.

Codex CLI verified

在 config.toml 新增 provider，Key 直接放进 http_headers 的 api-key。Add a provider in config.toml; put your Key directly into http_headers → api-key.

① 编辑配置文件① Edit the config file

路径Path

~/.codex/config.toml
Windows: C:\Users\<you>\.codex\config.toml

② 配置 provider + 默认模型② Provider + default model

config.toml

model = "YOUR_MODEL"
model_provider = "thetoken"

[model_providers.thetoken]
name = "The Token"
base_url = "https://api.thetoken.biz/openai/v1"
wire_api = "responses"                       # or "chat"
http_headers = { "api-key" = "YOUR_KEY" }

③ 验证③ Verify

shell

codex exec --skip-git-repo-check "What is 2+2?"

wire_api 用 responses（Codex 原生）或 chat 都能跑。wire_api can be responses (Codex native) or chat.

桌面聊天客户端Desktop chat apps

官方 ChatGPT 桌面版不支持自定义 API 端点（只能登录 OpenAI 账号）。请改用支持 “OpenAI 兼容端点” 的客户端：Chatbox / Cherry Studio / Jan / NextChat / LM Studio。The official ChatGPT desktop app does not support custom API endpoints (account login only). Use an OpenAI-compatible client instead: Chatbox / Cherry Studio / Jan / NextChat / LM Studio.

通用设置（字段名各客户端略有差异）Common settings (field names vary by app)

提供方 / API 类型Provider / API type

OpenAI / OpenAI-Compatible

API Host / Base URL

https://api.thetoken.biz/openai/v1

API Key

YOUR_KEY

模型名Model name

YOUR_MODEL （与卡片一致）(as on your card)

Chatbox — 设置 → 模型提供方选 OpenAI API → API 域名填 https://api.thetoken.biz/openai/v1 → 密钥填 YOUR_KEY → 模型填 YOUR_MODEL。Settings → provider = OpenAI API → API host = https://api.thetoken.biz/openai/v1 → key = YOUR_KEY → model = YOUR_MODEL.

Cherry Studio — 设置 → 模型服务 → 添加 OpenAI 提供商 → API 地址 https://api.thetoken.biz/openai/v1 → 密钥 YOUR_KEY → 添加模型 YOUR_MODEL。Settings → Model providers → add OpenAI → API host = https://api.thetoken.biz/openai/v1 → key = YOUR_KEY → add model YOUR_MODEL.

SDK / REST

Python （openai SDK）(openai SDK)

python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.thetoken.biz/openai/v1",
    api_key="YOUR_KEY",
)
r = client.chat.completions.create(
    model="YOUR_MODEL",
    messages=[{"role": "user", "content": "hi"}],
)
print(r.choices[0].message.content)

curl （REST）(REST)

shell

curl https://api.thetoken.biz/openai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"YOUR_MODEL","messages":[{"role":"user","content":"hi"}]}'

常见问题FAQ

现象Symptom	原因 / 处理Cause / fix
`403 model_not_allowed`	请求的 model 与 Key 绑定的模型不一致（用卡片上的 Model）The requested model differs from the one bound to your key (use the Model on your card)
`403 quota_exhausted`	额度用尽（含少量超额后）→ 联系管理员充值Quota used up (incl. small overage) → ask the admin to top up
`403 key_expired`	Key 到期 → 联系管理员续期Key expired → ask the admin to renew
`401 missing_api_key`	没带 Key，检查 Authorization 头No key sent — check the Authorization header
客户端 404Client 404	Base URL 必须以 /openai/v1 结尾Base URL must end with /openai/v1
agent 反复重试 / 额度飞快下降Agent retries / quota drains fast	Key 额度太小或算力不足；确认额度充足Key quota too small or throughput limited; ensure enough quota