Skip to main content
Virtual Keys provide a secure way to control access to LLM providers and models without exposing your actual provider API keys. They act as an abstraction layer that allows you to manage access permissions, rate limits, and model restrictions for different applications or users. Virtual Keys Overview The Virtual Keys page displays all configured virtual keys in a table format, showing their names, secret keys, allowed providers, models, rate limits, and creation dates.

​
What are Virtual Keys?

Virtual Keys are unique secret keys (prefixed with sk-uno-) that you can use instead of provider API keys in your applications. When a request comes in with a virtual key:
  1. The gateway validates the virtual key
  2. Checks which providers and models are allowed
  3. Enforces rate limits if configured
  4. Routes the request to the appropriate provider using your configured API keys
Benefits:
  • Security: Your provider API keys remain hidden from end users
  • Access Control: Restrict which providers and models can be accessed
  • Rate Limiting: Control usage with configurable rate limits
  • Flexibility: Different keys for different applications or environments
  • Monitoring: Track usage per virtual key

​
Creating a Virtual Key

To create a new virtual key:
  1. Navigate to the Virtual Keys page in the dashboard
  2. Click the Add Virtual Key button
  3. Fill in the form:
    • Name: Enter a descriptive name (e.g., “Production Access”, “Development Access”)
    • Providers: Select one or more providers that this key can access (required)
    • Models (Optional): Select specific models, or leave empty to allow all models
    • Rate Limits (Optional): Configure rate limits (see below)
  4. Click Create
Add Virtual Key Dialog The virtual key will be automatically generated and displayed. Make sure to copy and save it securely, as it won’t be shown again in full.

​
Selecting Providers

You must select at least one provider for the virtual key. Available providers include:
  • OpenAI - GPT models (GPT-4, GPT-3.5, etc.)
  • Anthropic - Claude models
  • Gemini - Google’s Gemini models
  • xAI - Grok models
  • Ollama - Self-hosted models
You can select multiple providers to allow the virtual key to access any of them.

​
Selecting Models

The Models field is optional. If left empty, the virtual key will have access to all models from the selected providers. If you want to restrict access to specific models:
  1. Click the Models dropdown
  2. Select the specific models you want to allow
  3. You can select multiple models
Note: Leaving models empty provides the most flexibility, as new models from the selected providers will automatically be accessible.

​
Configuring Rate Limits

Rate limits allow you to control how many requests can be made with a virtual key within a specific time period. You can configure multiple rate limits for different time windows. Rate Limit Configuration To add a rate limit:
  1. Click the Add Limit button in the Rate Limits section
  2. Select a Time Unit:
    • 1 minute
    • 1 hour
    • 6 hours
    • 12 hours
    • 1 day
    • 1 week
    • 1 month
  3. Enter the Limit (number of requests allowed in that time period)
  4. Click Add Limit again to add additional rate limits
Example Rate Limit Configurations:
  • 10 requests per 1 minute - Prevents burst traffic
  • 100 requests per 1 hour - Controls hourly usage
  • 1000 requests per 1 day - Daily quota management
You can combine multiple rate limits. For example:
  • 10/1min and 100/1h - Allows 10 requests per minute, but no more than 100 per hour
When a rate limit is exceeded, the gateway will return an error and reject the request.

​
Using Virtual Keys

Once created, you can use virtual keys in your applications just like provider API keys: 
# Using curl
curl -X POST https://your-gateway.com/api/gateway/openai/v1/chat/completions \
  -H "Authorization: Bearer sk-uno-your-virtual-key-here" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-4", "messages": [...]}'

# Using Python OpenAI SDK
from openai import OpenAI

client = OpenAI(
    base_url="https://your-gateway.com/api/gateway/openai",
    api_key="sk-uno-your-virtual-key-here"
)

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}]
)
The virtual key will be validated, and if allowed, the request will be routed to the appropriate provider using your configured API keys.