Custom Integrations
Pro / Enterprise
Connect your own AI endpoints to PromptPit. This is perfect for testing self-hosted models, corporate APIs, or fine-tuned models.
Supported Integration Types
OpenAI-Compatible
For endpoints that follow the OpenAI API specification:
- - Ollama
- - LM Studio
- - vLLM
- - Azure OpenAI
- - LocalAI
Custom HTTP
For any REST API endpoint:
- - Custom request body templates
- - JSONPath response parsing
- - Custom headers
- - Any authentication method
Setting Up an Integration
- Go to Settings > Integrations
- Click "Add Integration"
- Select integration type (OpenAI-Compatible or Custom HTTP)
- Enter connection details:
- Base URL of your endpoint
- Model identifier
- API key (if required)
- Optionally configure cost tracking
- Test the connection
- Save the integration
Integration Setup Form
[Image: Integration form showing fields for name, type selection, base URL, model identifier, API key input, and test connection button]
OpenAI-Compatible Configuration
# Example for Ollama
Base URL: http://localhost:11434/v1
Model: llama3.2
API Key: (leave empty)
Custom HTTP Configuration
For custom HTTP integrations, you can use template variables in your request body:
{
"prompt": "",
"max_tokens": ,
"temperature":
}Available template variables:
- The user's prompt text- Maximum tokens for response- Temperature setting- Your API key
Health Monitoring
PromptPit automatically monitors your custom integrations:
- Health checks run every 6 hours
- Integrations are automatically disabled after 5 consecutive failures
- You'll receive an email notification when an integration is disabled
- Re-enable manually after fixing the issue
Integration Limits
| Tier | Max Integrations |
|---|---|
| Free | Not available |
| Pro Basic | 2 integrations |
| Pro Plus | 5 integrations |
| Enterprise | 100 integrations |