OpenAI 相容的 API 閘道。改一行 base_url，沿用你現有的 SDK 與程式碼，不必重寫。多供應商路由、透明計費、按 token 實時扣減。
An OpenAI-compatible API gateway. Change one base_url, keep your existing SDK and code. Multi-provider routing, transparent per-token billing.
from openai import OpenAI client = OpenAI( base_url="https://26j.cn/v1", api_key="$API_KEY", ) r = client.chat.completions.create( model="gpt-5", messages=[{"role": "user", "content": "hello"}], ) # 其餘程式碼不需修改
每家供應商的認證方式、錯誤碼、限流規則、計費幣種都不一樣。這些工作不產生任何產品價值，但你每個月都得做一次。
Auth flows, error codes, rate limits, billing currencies — all different. None of it creates product value, but you maintain it every month.
沒有月費、沒有最低消費、沒有座位數。下表為每百萬 token 單價，實際扣減以控制台明細為準。
No subscription, no minimum, no seats. Prices below are per 1M tokens; the console statement is authoritative.
| Model | 輸入 / 1M | Input / 1M | 輸出 / 1M | Output / 1M | 計費 | Mode |
|---|---|---|---|---|---|---|
| gpt-5 | $0.110 | $0.800 | Token | Token | ||
| gpt-5-mini | $0.020 | $0.189 | Token | Token | ||
| claude-opus-4-5 | $0.177 | $0.880 | Token | Token | ||
| claude-sonnet-4-5 | $0.100 | $0.500 | Token | Token | ||
| gemini-2.5-pro | $0.240 | $2.000 | Token | Token | ||
| deepseek-v3.2 | $0.306 | $0.590 | Token | Token | ||
| kimi-k2.5 | $0.007 | $0.152 | Token | Token |
完整價格請見控制台模型列表。圖像與影片模型按次或按秒計費，規則不同，請於對應文件頁確認。價格如有調整，會提前於站內公告。
Full pricing lives in the console model list. Image and video models bill per request or per second — see their doc pages. Price changes are announced in advance.
模型清單隨時增減，請以控制台為準。所有模型共用同一組認證方式與回應格式。
Model availability changes; the console is authoritative. All models share the same auth and response format.
註冊後在控制台建立。建議每個環境用一把，方便單獨撤銷。
Create one in the console. Use a separate key per environment so you can revoke independently.
https://26j.cn/v1
請求與回應格式不變。在控制台查看延遲、用量與錯誤分佈。
Request and response shapes are unchanged. Watch latency, usage and error breakdown in the console.
如果你的程式用 OpenAI 官方 SDK 或相容客戶端，只需要換 base_url 與 Key。少數特殊參數（如部分供應商專屬欄位）可能不支援，會在文件標明。
If you use the official OpenAI SDK or a compatible client, you only swap base_url and the key. A few provider-specific params aren't supported and are documented.
按實際 token 用量扣減，用多少扣多少，沒有月費。控制台可即時查看每筆請求的用量明細。建議設定餘額提醒與單 Key 限額。
Per actual token usage. No subscription. The console shows per-request detail in real time. Set a balance alert and per-key limits.
支援 SSE 串流，OpenAI 相容格式。stream: true 即可，無需額外設定。
Yes — SSE streaming in the OpenAI-compatible format. Just set stream: true.
請求內容會轉發至對應的上游模型供應商以完成推論。平台記錄呼叫時間、用量與計費資料用於結算與故障排查。若你的業務有資料不出境或強隱私要求，請勿使用本服務。
Prompts are forwarded to the corresponding upstream provider to produce a completion. We log timing, usage and billing for settlement and debugging. If you have strict data-residency requirements, do not use this service.
可以，請於控制台提交開票申請，或聯繫客服。
Yes — request one in the console, or contact support.
未消耗餘額在符合條款條件下可申請退款，詳見服務條款第三條。因程式錯誤、密鑰洩漏或反覆重試造成的消耗不在退款範圍。
Unused balance is refundable subject to the terms (see §3). Consumption from code bugs, leaked keys or retry loops is excluded.
註冊即送試用額度，不需要綁定信用卡。
Free credits on signup. No card required.