01 / Development pipeline
One API key. 37 text models.
A unified gateway, 100% compatible with the OpenAI protocol. Change only base_url and api_key — your existing SDKs, workflows and agent stack need no modification.
Why teams choose us
Drop-in replacement, zero migration cost
Fully standard OpenAI REST format. Change base_url and api_key — no new SDK, no refactoring, no re-testing the whole chain.
Chinese models at 0.6× official price
DeepSeek, GLM, Qwen, Kimi, MiniMax and Hunyuan are all billed at 0.6× the official rate. Prices come from a public endpoint anyone can re-check.
Works with the ecosystem out of the box
Cursor, NextChat, Chatbox, LangChain, LlamaIndex, and any application that accepts a custom API host — fill it in and it works.
Prices you can verify
The model catalogue and multipliers come from an unauthenticated public endpoint. We do not hide multipliers or quote pricing only on request.
Change two lines. Start calling.
Any OpenAI SDK can point at our gateway directly. Here is the minimal working example.
from openai import OpenAI
client = OpenAI(
api_key="your_alphatech_key",
base_url="https://alphatech.net.cn/v1",
)
response = client.chat.completions.create(
model="deepseek-v4.1-flash",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)base_url must include the /v1 suffix. Keys are generated in the console; you can issue multiple keys per project with separate quotas.
Base URL:https://alphatech.net.cn/v1
Models and prices currently live
Billed by usage, in CNY per million tokens. Prices follow upstream changes; the console shows the real-time figure.
| Model | Group | Input / 1M | Output / 1M |
|---|---|---|---|
| gpt-6-luna | default | ¥0.21 | ¥1.05 |
| deepseek-v4.1-flash | cn | ¥0.6 | ¥2.4 |
| qwen3.8-flash | cn | ¥0.8 | ¥2.7 |
| glm-5.3-flash | cn | ¥0.8 | ¥2.8 |
| gpt-5.6-sol | default | ¥8.4 | ¥42 |
| claude-new-opus-5 | default | ¥10.5 | ¥68.25 |
The table above is a sample. For the complete catalogue and live multipliers, see the console or the public pricing endpoint.
Public pricing endpoint →See the full model catalogue →Supported clients
Anything that lets you set a custom API base URL will work. These are the most common ones in the community.
Operating boundaries
Billing
Input and output tokens are billed separately. Failed requests are not billed. Multipliers are shown in the console.
Rate limits
The default group is limited by concurrency and requests per minute. A low-speed group is available for individual trials. Enterprise terms are negotiated separately.
Data boundary
The gateway forwards requests and does not use request content for model training. Log retention scope and period are described in the privacy policy.