AI Proxy Capabilities

Overview

The AI Proxy provides powerful capabilities that work seamlessly across multiple AI providers, enabling you to build sophisticated AI applications without being locked into a single provider's ecosystem.

Core Capabilities

🔧 Tool Calling

Enable AI models to execute functions and interact with external systems.

Function Execution: Call predefined functions with structured parameters
Parallel Execution: Run multiple tools simultaneously for efficiency
Streaming Support: Stream tool calls and arguments in real-time
Provider Support: Works across OpenAI, Anthropic, Google AI, and more

Use Cases: API integration, database queries, calculations, external service calls

👁️ Vision

Analyze and understand images across multiple formats and providers.

Multi-format Support: JPG, PNG, WEBP, GIF, HEIC
Batch Processing: Analyze multiple images simultaneously
Detail Control: Adjust processing quality and cost
OCR Capabilities: Extract text from images and documents

Use Cases: Document analysis, content moderation, accessibility, medical imaging

📊 Structured Outputs

Generate guaranteed JSON responses that conform to your schemas.

JSON Schema Validation: Ensure response format compliance
Strict Mode: Guarantee schema adherence with no parsing errors
Complex Structures: Support nested objects, arrays, and custom types
Real-time Validation: Validate responses as they're generated

Use Cases: Data extraction, API responses, form processing, database integration

🧠 Reasoning

Access advanced reasoning models with transparent thinking processes.

Chain of Thought: See step-by-step reasoning and problem-solving
Thinking Budget: Control computational resources for reasoning
Multiple Providers: OpenAI o1/o3, Claude thinking, Gemini reasoning
Streaming Reasoning: Watch the reasoning process unfold in real-time

Use Cases: Complex problem-solving, mathematical computations, research analysis, tutoring

📄 PDF Input

Process PDF documents for content extraction and analysis.

Large File Support: Handle documents up to 500MB
Text Extraction: Extract and understand document structure
OCR Processing: Handle scanned documents and images
Multi-page Analysis: Process complex documents with multiple pages

Use Cases: Document analysis, contract processing, research paper analysis, compliance checking

🌊 Streaming

Real-time token-by-token response generation for better user experience.

Server-Sent Events: Standard SSE streaming across all providers
Tool Call Streaming: Stream function arguments as they're generated
Multi-provider: Consistent streaming interface across providers
Error Recovery: Robust handling of network interruptions

Use Cases: Chat applications, content generation, real-time analysis, interactive experiences

Provider Compatibility Matrix

Capability	OpenAI	Anthropic	Google AI	AWS Bedrock	Azure	Groq	Cohere
Tool Calling	✅	✅	✅	✅	✅	✅	✅
Vision	✅	✅	✅	✅	✅	❌	❌
Structured Outputs	✅	✅	✅	✅	✅	✅	✅
Reasoning	✅	✅	✅	❌	❌	❌	❌
PDF Input	❌	✅	✅	✅	❌	❌	❌
Streaming	✅	✅	✅	✅	✅	✅	✅

Getting Started

Quick Setup

All capabilities are available through the same unified API endpoint:

<CODE_PLACEHOLDER>

Authentication

Use your ORQ API key for all capability requests:

<CODE_PLACEHOLDER>

Basic Usage Pattern

Each capability follows a consistent request format:

<CODE_PLACEHOLDER>

Combining Capabilities

Multi-Modal Analysis

Combine vision and structured outputs for image analysis:

<CODE_PLACEHOLDER>

Reasoning with Tools

Use reasoning models with function calling:

<CODE_PLACEHOLDER>

Streaming Structured Outputs

Stream structured responses in real-time:

<CODE_PLACEHOLDER>

Best Practices

Performance Optimization

Provider Selection: Choose providers based on specific capability strengths
Request Batching: Combine multiple operations when possible
Caching: Cache responses for repeated requests
Error Handling: Implement robust fallback strategies

Cost Management

Token Monitoring: Track token usage across different capabilities
Provider Comparison: Compare costs across providers for your use case
Optimization: Use appropriate detail levels and processing options
Budget Controls: Set limits on reasoning token budgets and file sizes

Security Considerations

API Key Protection: Secure your ORQ API keys
Input Validation: Validate all inputs before processing
Content Filtering: Implement appropriate content moderation
Data Privacy: Handle sensitive data according to your privacy requirements

Integration Examples

Node.js SDK

<CODE_PLACEHOLDER>

Python SDK

<CODE_PLACEHOLDER>

REST API

<CODE_PLACEHOLDER>

Support and Resources

Documentation

API Reference: Complete API documentation
SDKs: Official SDK documentation
Examples: Implementation examples and tutorials

Community

GitHub: Report issues and contribute to open-source SDKs
Discord: Join the developer community for support
Support: Contact support for enterprise assistance

Advanced Features

Retries & Error Handling: Robust error recovery
Load Balancing: Distribute requests across providers
Monitoring: Track performance and usage metrics

Next Steps

Choose Your Capability: Select the AI capability that matches your use case
Review Provider Support: Check which providers support your needed features
Start with Examples: Use our code examples to get started quickly
Scale Gradually: Begin with simple implementations and add complexity over time
Monitor Performance: Track usage, costs, and performance as you scale