Model capabilities keep expanding across China and global access
China access aggregates leading local models, while global access connects leading international models.
Access general-purpose, image, video, speech, and vector models through one stable API with unified metering, calls, and management.
China access aggregates leading local models, while global access connects leading international models.
Support for formal contracts, corporate payments, invoices, verified model sources, and auditable usage.
Review pricing, balances, usage, and API key limits in one operating view.
Discuss data residency, edge collaboration, and industry delivery with the solution team.
Browse the current site catalog with public prices and capabilities, then choose the right model in the console.
Browse all modelsChoose the service model that fits your workload, resource isolation, and data boundary.
Standard API, pay for actual usage
For fast validation, elastic workloads, and unified developer access.
Open consoleDedicated resources and operating boundaries
For stable production calls, resource isolation, and dedicated business service.
Explore dedicated accountsRun model services in a designated environment
For sensitive data, private networks, and industry delivery.
Explore private deploymentSupport formal contracts, corporate payments, and invoices.
Connect to verified upstream model services.
Review usage by customer, model, and API key.
Set balances and limits by customer and API key to keep spending visible.
MaaSLink is a model-service platform for enterprises, development teams, and channel partners. It brings text, image, audio, video, and vector models into one service, so teams can create API keys, make calls, and review usage and quotas without learning a separate workflow for every provider.
Before choosing a model, open the model catalog to see what is currently available, which inputs and outputs it supports, and the price shown for the current site. Then follow the online documentation to complete a test integration before moving the workload into production. This keeps your test configuration aligned with the model you plan to use.
Model availability, features, and prices can change as the catalog is updated, so use the information shown on the site when you make a decision. To learn more about the platform's background and service focus, visit About us.
To make a first MaaSLink model call, create an account and an API key, confirm the Base URL for the current site, and copy a model ID from the model catalog. Keep the API key in a server-side secret manager; do not place it in browser code, a public repository, or application logs.
Next, follow the online documentation to configure a compatible client or send the API request directly. Check that the response matches the input and output requirements of the model you selected, and run the whole process in a test environment before using it in production.
If a call fails, check the Base URL, model ID, account balance or quota, authentication header, and request format in that order. Model prices, rate limits, and supported features can change, so confirm the current catalog entry before launch. The complete setup steps and common troubleshooting guidance are available in the Documentation section.
A channel partner can use MaaSLink to link approved customer accounts, set balances or limits within its authorized scope, and review usage by customer, model, or API key. This keeps customer operations, credentials, and metering in one workflow instead of requiring every team to build its own management system.
Before delivering a service, confirm the customer relationship and the permissions available to the account. Quotas, settlement, and service scope should follow the agreement with the customer, while the models and features that can be used depend on the current account and catalog.
Review each customer link and quota change in the console before it takes effect. The channel workflow and step-by-step instructions are available in Documentation, and current model details and prices are listed in the model catalog. Check both before confirming a customer deployment.
From qualification review to customer affiliation, credit allocation, and wallet records, partners can run model service operations from one console.
Provide enterprise, contact, and qualification details.
Track the review status and provide additional materials when requested.
Process requests from customers who want to join the partner account.
Transfer available account credits to affiliated customer accounts.
Use the partner dashboard and wallet ledger to follow customer count, pending requests, and recent credit movements.
Register and activate an individual or business account.
Generate a key in the console.
Compatible with the OpenAI SDK.
Select the model that fits your use case.
Compatible with the OpenAI SDK calling pattern.
Connect leading AI coding tools.
Connect common desktop assistants through a compatible API endpoint.
Use model access in knowledge bases, assistants, workflows, and low-code AI applications.
After registration, follow the journey for your account type. Enterprise users can configure calls directly; channel users first submit a channel application.
Choose the interface that matches the client: Cursor uses OpenAI Chat Completions, Claude Code uses Anthropic Messages, and Codex uses OpenAI Responses. Follow each tool's own fields for the address, key, and model ID shown on this page.
These answers help you quickly resolve common setup and account questions.
Review the terms and agreements that apply to your use of MaaSLink services.
These answers help you quickly resolve common setup and account questions.
Sign in to the console and create an API key in key management. Use the key only in controlled environments; never put it in client-side code, public repositories, or screenshots.
Use the address shown on this page for the site you are visiting. OpenAI-compatible calls, Cursor, OpenClaw, and Hermes Agent use the URL ending in `/v1`; Codex uses OpenAI Responses and must also retain `/v1`; Claude Code's `ANTHROPIC_BASE_URL` omits `/v1` because the client appends `/v1/messages`. Do not mix credentials or addresses between sites.
Use a model ID currently available in the model hub or console for this site. Availability can change, so do not use `MODEL_ID` from the example directly in production.
For server-side requests, send the API key as `Authorization: Bearer YOUR_API_KEY`. When a tool uses its own authentication fields, map them according to that tool's official documentation.
You can view application status and add requested materials. Customer association, quota configuration, and usage operations become available only after approval.
Create an account to try unified model access.