Get an API Key and Authenticate

Before you start calling the API, please first learn how to obtain and manage your dedicated key.

1. AI Large Model API Key (for calling LLMs, image generation, and video models)

This key is used to access all the generative AI models we provide in the “Model Plaza”.

1.1 Obtain and Create Tokens (Token)

• 1. Log in to the console, find and click API key in the left navigation bar (located below "Usage Logs").
• 2. For better security control, we recommend clicking "+ Add Token" at the top of the page to generate separate keys for different projects.

1.2 Core Configuration: Group Types and Token Groups

When creating a new token, you need to choose a group type. The system supports three intelligent routing strategies:

• Cost first: dynamically routes to the underlying channel with the best current cost-performance ratio.
• Stability first: dynamically routes to the channel with the highest current availability (high SLA).
• Specified group: manually configure underlying channels with fine-grained control (workflow: select "Model Provider" -> select "Token Group").Under "Specify Group", group names are typically composed of a prefix (model brand or channel source) + a suffix (SLA attributes): Common prefix examples (representing model brands or cloud channels):
• Text and core large models
  • gpt / openai: OpenAI series
  • claude / cl / aws: Anthropic Claude series and underlying cloud provider channels
  • gemini / gm: Google Gemini series
  • deepseek / ds: DeepSeek series

Core Suffix Guide (indicating underlying SLA and concurrency characteristics):

• -of (Official / Official Channel): 100% official native access or direct connection to leading cloud providers. Extremely high SLA, the preferred choice for enterprise production environments.
• -vip (VIP Pool / High-Availability Pool): High-quality large-concurrency pool or dedicated independent nodes. Ensures high stability under heavy concurrency.
• -sp (Special Pool / Value Pool): Standard value resource pool. Extremely cost-effective, the preferred choice for development testing and everyday code debugging.
• -s (Saver / Spot-Discount Pool): Ultra-low-cost resources (stability not guaranteed), suitable for tasks that are extremely cost-sensitive.
• -sl / -fa (Slow & Fast): Slow / fast channels. Used specifically for speed tiers in image generation tasks such as Midjourney.
• -pro / -cli / -re: Exclusive hybrid resource pools for specific account systems or specific scenarios.

⚠️ Important Notice: Because the system dynamically provisions available resources based on the current account’s permission level and business scenario, the length of group names and the exact prefixes shown may vary slightly across accounts (for example, Claude may appear as claude-aws-of or abbreviated as cl-of). Please always refer to the actual group name displayed in your current console dropdown list. If you have requirements for enterprise-grade high concurrency or customized private deployment, please contact your dedicated business representative.

2. Key Security and Advanced Configuration

When creating a new token, APIShop provides a set of enterprise-grade risk control options, and we strongly recommend configuring them before going into production:

• Quota Limit: By default, tokens have no quota limit (subject to your account’s total balance). You can set a fixed total USD amount for tokens issued to team members or specific tools. Once the limit is exceeded, the token becomes invalid, effectively preventing financial losses caused by runaway code loops.
• Expiration: Set it with one click to "one month," "one day," or a custom date and time, making it ideal for short-term outsourcing projects or temporary testing.
• Access Restriction (IP Whitelist): Supports binding specific server IP addresses. Even if a key is accidentally leaked, IPs outside the whitelist cannot make requests, fundamentally preventing malicious abuse and unauthorized charges.