LLM APIs for universities and research institutes
Research groups use it to compare models from several vendors, search literature and analyze text; teaching teams write questions and get help with grading; university IT builds campus Q&A. One account and one API key reach models from many vendors, billed by actual usage with no monthly fee or minimum spend. Each student can have a key with its own spending limit, and we issue China domestic invoices (fapiao) and overseas invoices for grant reimbursement.
Research
Multi-model experiments, model evaluation, literature reading, social-science text analysis and data labeling. The same code compares models by changing the model parameter, and every call's tokens and cost are recorded.
Compare several models on the same question set: recommended models, a reproducible logging method and a monthly budget estimate.
Solution
Score models in bulk on your own evaluation set and compare quality, speed and cost.
Solution
Classify and code interview transcripts, open-ended survey answers and online text.
Solution
Read papers in bulk, pull out key points and compare findings to gather material for a review. It helps you read and organize, not write.
Solution
A cheap small model labels in bulk, low-confidence items go to a second vendor's model and a sample is checked by people.
Solution
Turn the group's papers, lab notes and protocols into a searchable library you can ask questions.
Solution
Teaching
Writing questions, grading help and course Q&A, with a teacher approving every question before it enters the bank and every grade before it is published.
One model writes questions from the textbook and another vendor's model answers them independently to check the key; rubric-based grading help and a course assistant that answers only from course material.
Solution
Campus services
University IT turns rules, procedures and FAQs into a help service that hands off to a person when it cannot answer.
A small model sorts each question, the answer is drafted from matching knowledge-base entries, and hard cases go to a person.
Solution
Rules and how-to guides as a knowledge base, answered with embeddings, rerank and a chat model.
Solution
Running a lab account
An account has no member sub-accounts, so a lab splits it into keys per student and project: limits, expiry, usage and reconciliation all work per key.
The PI holds one account, each student gets a key, and spending is booked and reimbursed per key.
Learn
One key per member, project and environment, each with its own spending limit, expiry, model limits and IP allowlist.
Learn
Sum the spend records exported from Logs by key and subtract refunds to get each project's cost.
Learn
Keep keys on the server, out of repositories and screenshots, and give every key a spending limit so a leak stays small.
Learn
Buying and invoices
Pay as you go with no monthly fee or minimum spend. We issue both China domestic invoices (fapiao) and overseas invoices: add our WeChat support (JsnonoChat) and send your order number.
¥1 buys ¥1 of credit, charged by actual usage; chat models bill by tokens, image models and similar per request.
Page
Buying for a school or department: purchase redemption codes offline through support and redeem them for credit in the console, which suits bulk top-ups.
Page
Invoices, bulk purchases and integration questions: add our WeChat support JsnonoChat.
Page
How a lab gets started
Sign up, top up, hand out keys and try models: four steps before students can use it.
01
Create the account
The PI or grant administrator signs up for one account and turns on two-step verification. Only the owner signs in; the password is never shared.
02
Top up
Top up online on the console's Top up page (as shown in the console), or buy redemption codes offline through support. Usage is charged as it happens; ¥1 buys ¥1 of credit.
03
One key per student
On the API keys page, create a key per student or project with a spending limit and an expiry; you can create up to 100 at once. All keys draw on the account balance.
04
Try models before writing code
Once signed in, try a few models on your own questions in the playground, charged at the chosen group's price, or start with the model comparisons.
FAQ
Can we get invoices for grant reimbursement?
Yes. We issue both China domestic invoices (fapiao) and overseas invoices: add our WeChat support (JsnonoChat) and send your order number. Copy the order number from All records on the console's Top up page.
Can each student have their own key?
Yes. An account has no member sub-accounts, so a lab splits it with API keys: on the console's API keys page, create one key per student with its own spending limit and expiry; you can create up to 100 at once. The limit is cumulative and does not reset monthly, and all keys draw on the account balance. Filter Logs by key name to see each key's usage and export it as CSV. See teams and multiple projects on one account.
Does the platform keep the research data we send to models?
Under the default configuration, usage logs record one entry per call: time, model, group, key name, token counts, cost and latency. They do not record request or response bodies, nor the caller's IP address. Asynchronous jobs such as video generation store the task status and returned result so you can check progress and fetch the output. To generate a response, a request is forwarded to the model provider behind the group you chose, and how that provider handles data is governed by its own policies; if the source matters, choose an Official or Cloud group. Details are on security and compliance.
How do we choose a model?
Start from task difficulty: small, cheap models for classification and extraction, mid-tier models for Q&A and summaries, flagship or reasoning models for hard multi-step work, then test on a few dozen items of your own data. The model comparisons put common models side by side with each group's price, input and output types and supported endpoints; models and pricing lists every model's price in every group.
How does billing work? Is there a monthly fee?
No monthly fee and no minimum spend commitment. You top up first and usage is charged as it happens, with ¥1 buying ¥1 of credit: chat models bill by tokens, with input and output priced separately; image models and similar bill per request. Each call is charged at the price in effect when it is made and recorded in Logs in the console.
What happens when a student graduates or leaves the group?
Disable the keys they used. Disabling beats deleting: a disabled key can be turned back on and still shows by name on the Dashboard. If they had access to a key the group shares, create a new one, switch to it, then disable the old one.
Sign up, then give each student a key
After signing up, create keys with spending limits on the API keys page and put the base_url and key into your code. For school-wide purchasing or invoices, add our WeChat support JsnonoChat.
Scan the code or search the WeChat ID to add us