
Shuyan AI offers an enterprise model...
Products that both call a model and turn web pages into usable text usually stitch two vendors together. Here, page parsing emits quotable text and structured fields, web search adds live results, and documents or images go through text recognition. After you point model calls at the platform address and key, the console shows token spend and can limit models and quota per group. Agent workloads use the same token billing for compute and quota, rather than a second vendor that only sells chat. It fits development teams that need both a model API and structured pages or documents. You must register and either use the new-user token grant or add balance; with no balance the API stops returning results. Even though chats are not stored, you still configure key permissions and groups yourself.
Shuyan AI offers an enterprise model cloud and data APIs. On the model side, one base address calls text, audio, image, and video, using the common programming interface, and keys can be limited by quota and by which models they may use. On the data side, page parsing, web search, search add-ons, and document text recognition turn pages or files into structured results. New accounts receive tokens to try the service, then calls draw on the account balance. The site says conversation content is not stored and traffic is encrypted.
Difficulty: Intermediate
One model base address
Call text, audio, image, and video models from one base address, using the common programming interface and developer tools.
Parse pages into structure
Turn a web page into text and structured fields a model can use, aimed at complex pages rather than a single plain-text dump.
Web search and add-ons
Offers retrieval-based web search and search add-ons with live data so an app can fetch outside results before answering.
Document and image text recognition
Extract structured content from documents and images for knowledge preparation or a later model step.
Keys, groups, and quota
Create keys by user group, set quota and allowed models separately, and review token spend in the console.
One gateway for app models
A developer points text, image, or video calls at the Shuyan address and uses separate keys to limit models per environment.
Turning pages into source text
When collecting a docs page or product page, call page parsing and send the structured result to the model instead of pasting the whole page.
Extracting fields from scans
Send a contract image or exported document through text recognition, take the title and body blocks, then continue into search or review.
No. Besides the model API, the product page lists page parsing, web search, search add-ons, and document text recognition. Models use one address for text, audio, image, and video and follow the common programming interface. The data APIs turn pages or documents into structured results. Both use the same account and keys, but confirm that the current key is allowed to call each one.
The page says new users receive tokens so they can send a first request; ongoing use means adding balance and calling with a key. Keys can set quota and allowed models per group. After the grant is gone, calls stop if there is no balance. The grant size and the model list change, so use the account page after you register.
The security section says traffic is encrypted end to end and conversation content is not stored. It also mentions audit trails and environment isolation. That does not replace your own key policy: still limit quota, allowed models, and who can see spend, by group. If compliance needs more, read the privacy terms rather than the homepage slogan alone.
Real reviews and feedback from users