Skip to content

SJTU (致远一号 AI 模型 API)

For Shanghai Jiao Tong University affiliates only. SJTU's High-Performance Computing Center runs an on-campus LLM gateway at models.sjtu.edu.cn that hosts several leading Chinese / open-source models behind a single OpenAI-compatible endpoint. If you're an SJTU student or staff member, this is often the fastest path to experimenting with Chinese-ecosystem models without paying per-token — the test-phase quota is large and the API shape is the same as every other provider in this tutorial.

Sign up and get a key

  1. Log in to My SJTU (交我办) at https://my.sjtu.edu.cn/ — web or mobile app.
  2. Search for the workflow "致远一号" AI 模型 API 申请（测试） ("Zhiyuan-1" AI Model API Application — Test) and start an application.
  3. Fill out the form (contact email, which models you want, intended use) and submit.
  4. Wait for approval. Once granted, the base_url and api-key are emailed to you and also delivered to your Jiao-Wo-Ban inbox.

Keys are personal — selling or leaking the key cancels your access. The test-phase key expires 2026-06-30.

Quota (test phase)

  • 100 requests per minute
  • 100,000 tokens per minute
  • 1,000,000,000 tokens per week

That's an enormous allowance by individual-user standards — enough to run the entire tutorial's agent loops many times over. Larger campus applications (research groups, classroom deployments) can request a higher quota by emailing hpc@sjtu.edu.cn.

Install

pip install openai

Nothing SJTU-specific — the endpoint speaks the OpenAI wire format.

Hello world

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["SJTU_API_KEY"],
    base_url="https://models.sjtu.edu.cn/api/v1",
)

resp = client.chat.completions.create(
    model="deepseek-chat",
    messages=[{"role": "user", "content": "Say hi in one short sentence."}],
)
print(resp.choices[0].message.content)

Loading the key

SJTU_API_KEY=sk-...

Model catalog

Use the exact id returned by GET /api/v1/models, not the human-readable label you saw on the application form. Current catalog:

model id Underlying model Context Notes
deepseek-chat DeepSeek V3.2 (685B) 32k General-purpose. Default for the tutorial's notebooks.
deepseek-reasoner DeepSeek V3.2 reasoning 32k Returns reasoning_content alongside content.
glm-5 Zhipu GLM-5 (744B) 32k Strong on coding and agent tasks.
minimax / minimax-m2.5 MiniMax M2.5 (230B) 192k Use this when you need a long context.
qwen3coder Qwen3-Coder (30B) 32k Coding-focused.
qwen3vl Qwen3-VL (30B) 32k Vision / multimodal.

You can verify the live list any time with:

curl 'https://models.sjtu.edu.cn/api/v1/models' -H 'Authorization: Bearer your-api-key'

Gotchas

  • Network: the endpoint is only reachable from the SJTU campus network, or from off-campus via the SJTU VPN. Colab, Deepnote, or any non-SJTU cloud instance will timeout against this URL. For those environments, fall back to one of the public providers.
  • V3.2 requires a user message: if your messages list contains only a system entry with no user message, deepseek-chat and deepseek-reasoner return an empty response. Always include at least one {"role": "user", ...} turn. (The tutorial's code always does, so nothing to change — just worth knowing if you hand-roll a request.)
  • Key expiry: the test-phase key is valid through 2026-06-30. Renew when announced.
  • Use the id, not the form label: e.g. the form may show "DeepSeek-V3.2", but the API call needs model="deepseek-chat".

Further reading