Skip to main content

​
General Case: No Need to Set

Recommended Practice: When creating a token, there is no need to set available models, allowing you to use all models on the site.
Available models use a whitelist mechanism:
  • Not Set: The token can call all 200+ models on the site
  • Set: The token can only call specified models and cannot use others
If your business scenarios are diverse, setting available models will limit the token’s usage scope and affect flexibility.

​
Why Not Recommend Setting?

​
1. Limits Call Range

Suppose you set available models to gpt-4o, claude-3-5-sonnet-20241022, then:
  • ✅ Can call: gpt-4o, claude-3-5-sonnet-20241022
  • ❌ Cannot call: gemini-2.5-pro, deepseek-v3, other 200+ models
This will severely limit your business flexibility.

​
2. Model Alias Issues

Model Alias Trap: Some models have aliases, which can easily cause errors when setting available models.Case Study:
  • Alias: nano-banana-pro (friendly nickname)
  • Official Name: gemini-3-pro-image-preview (official API name)
If you set available models to nano-banana-pro, but call gemini-3-pro-image-preview in code, you’ll get an error:
{
  "error": {
    "message": "Model not allowed for this API key",
    "type": "invalid_request_error"
  }
}

​
3. Business Scenario Changes

As your business evolves, you may need to test or switch different models:
  • Switch from GPT-4 to Claude 3.5
  • Test domestic models like DeepSeek, Qwen
  • Try newly released Gemini 2.5
If you set available models in advance, you’ll need to modify the whitelist every time, which is very troublesome.

​
When Should You Set It?

Although not recommended, you can consider setting available models in the following scenarios:

​
Scenario 1: Sharing Tokens with Others

Team Collaboration/Friend Sharing

If you share a token KEY with colleagues or friends and want them to only use specific models (such as gpt-3.5-turbo) to avoid consuming your balance with expensive models, you can set available models.Example:
  • Your balance is limited, you only want colleagues to use gpt-3.5-turbo, gpt-4o-mini
  • After setting available models, even if colleagues try to call o3, they will be rejected

​
Scenario 2: Budget Control

Strict Cost Management

If you set a strict budget for a project and only allow using low-cost models, you can restrict through available model whitelist:Example:
  • Project budget is limited, only allow using deepseek-v3 ($0.07/million tokens)
  • Prohibit using o3 ($20/million tokens) to avoid overspending

​
Scenario 3: Security Compliance

Enterprise Security Policy

Some enterprises have strict AI model usage regulations, only allowing use of approved models.Example:
  • Company only allows using official OpenAI models
  • Set available models to gpt-4o, gpt-4.1, prohibit using other vendor models

​
How to Set Available Models?

If you really need to set available models, follow these steps:
  1. Log in to API.YI Token Management Page
  2. Click “Add New KEY” or edit existing token
  3. Select allowed models in the “Available Models” field
  4. Save configuration
Use Official Names: When setting available models, it’s recommended to use the official names of models (such as gemini-3-pro-image-preview) instead of aliases (such as nano-banana-pro) to avoid call failures.You can view all model official names on the Model List Page.

​
Summary

Setting MethodAdvantagesDisadvantagesApplicable Scenarios
Not Set Available Models✅ Can use all 200+ models
✅ No need to worry about alias issues
✅ High business flexibility		❌ Cannot restrict others from using expensive modelsPersonal use, development testing, production environment (recommended)
Set Available Models✅ Strictly control model usage scope
✅ Prevent expensive models from consuming balance		❌ Limits flexibility
❌ May encounter alias issues
❌ Need to manually maintain whitelist		Shared tokens, budget control, security compliance
Recommended Practice: Unless there is a clear control requirement, do not set available models and enjoy access to all models.