OpenAI Provider

OpenAI sets the industry standard for AI models, offering the most mature ecosystem and reliable performance for production applications. Through CCProxy integration with Claude Code, you can access GPT-4o, GPT-4.1, and the latest o3/o4 reasoning models while maintaining your familiar development workflow.

🏭 Why Choose OpenAI for Claude Code?

• 🥇 Industry standard: Most mature and reliable AI models with proven enterprise adoption
• 🛠️ Complete parameter support: Temperature, top_p, presence_penalty, frequency_penalty all work perfectly
• 👁️ Advanced vision: Best-in-class image understanding with GPT-4o and GPT-4.1
• 🎯 Function calling: Full support for Claude Code's tool use capabilities
• 🧠 Reasoning models: Access to o3 and o4-mini for complex problem-solving (text-only)
• 🔧 Zero-config integration: Works immediately with CCProxy, no modifications needed

Setup

1. Get an OpenAI API Key

1. Visit platform.openai.com
2. Sign up for an account
3. Navigate to the API Keys section
4. Generate a new API key

2. Configure CCProxy

Create or update your CCProxy configuration file:

mkdir -p ~/.ccproxy
cat > ~/.ccproxy/config.json << 'EOF'
{
  "providers": [
    {
      "name": "openai",
      "api_base_url": "https://api.openai.com/v1",
      "api_key": "sk-your_openai_api_key_here",
      "models": ["gpt-4o", "gpt-4o-mini", "gpt-4.1", "gpt-4.1-mini", "gpt-4.1-nano", "o3", "o3-pro", "o4-mini"],
      "enabled": true
    }
  ],
  "routes": {
    "default": {
      "provider": "openai",
      "model": "gpt-4o"
    }
  }
}
EOF

3. Start CCProxy and Claude Code

# Start CCProxy
ccproxy start

# Configure Claude Code to use CCProxy
ccproxy code

Available Models

OpenAI provides various model families optimized for different use cases:

Latest Models (July 2025)

Model	Context Length	Best For	Function Calling
gpt-4o	128K tokens	Flagship model with vision, most versatile	✅ Yes
gpt-4o-mini	128K tokens	Smaller, cost-efficient version of GPT-4o	✅ Yes
gpt-4.1	128K tokens	Specialized for coding tasks, precise instruction following	✅ Yes
gpt-4.1-mini	128K tokens	Smaller variant of GPT-4.1	✅ Yes
gpt-4.1-nano	128K tokens	Smallest, most efficient GPT-4.1 variant	✅ Yes
o3	128K tokens	Most powerful reasoning model, excels at complex problems	❌ No*
o3-pro	128K tokens	Extended thinking version of o3	❌ No*
o4-mini	128K tokens	Fast, cost-efficient reasoning model	❌ No*

Legacy Models

Model	Status	Notes
gpt-4-turbo	Available	Previous generation, consider GPT-4o
gpt-4	Available	Original GPT-4
gpt-3.5-turbo	Available	Budget option, fastest response times

⚠️ Important: Claude Code requires models with function calling support. Models marked with ✅ work perfectly with Claude Code.

*Note: o3/o4 reasoning models don't support function calling but can be used for pure text generation tasks when tool use is not required.

Configuration Options

Complete Configuration Example

{
  "providers": [
    {
      "name": "openai",
      "api_base_url": "https://api.openai.com/v1",
      "api_key": "sk-your_api_key",
      "models": ["gpt-4o", "gpt-4o-mini", "gpt-4.1", "gpt-4.1-mini", "gpt-4.1-nano", "o3", "o3-pro", "o4-mini"],
      "enabled": true,
      "timeout": "60s",
      "max_retries": 3,
      "organization": "org-your_org_id",
      "headers": {
        "OpenAI-Beta": "assistants=v2"
      }
    }
  ],
  "routes": {
    "default": {
      "provider": "openai",
      "model": "gpt-4o"
    },
    "longContext": {
      "provider": "openai",
      "model": "gpt-4o",
      "conditions": [{
        "type": "tokenCount",
        "operator": ">",
        "value": 60000
      }]
    },
    "background": {
      "provider": "openai",
      "model": "gpt-4.1-nano"
    },
    "think": {
      "provider": "openai",
      "model": "o3"
    }
  }
}

Using Environment Variables

CCProxy can automatically detect your OpenAI API key from environment variables:

# Option 1: Auto-detection (recommended)
export OPENAI_API_KEY="sk-..."

# Your config.json can omit the api_key entirely:
{
  "providers": [{
    "name": "openai",
    "api_base_url": "https://api.openai.com/v1",
    "enabled": true
    // api_key not needed - auto-detected from OPENAI_API_KEY
  }]
}

# Option 2: Variable substitution in config
{
  "providers": [{
    "name": "openai",
    "api_key": "${OPENAI_API_KEY}",
    ...
  }]
}

Advanced Features

Supported Parameters

OpenAI has the most complete parameter support in CCProxy:

{
  "model": "gpt-4o",
  "messages": [...],
  "temperature": 0.7,          // 0.0 to 2.0
  "top_p": 0.9,               // 0.0 to 1.0
  "presence_penalty": 0.0,     // -2.0 to 2.0
  "frequency_penalty": 0.0,    // -2.0 to 2.0
  "max_tokens": 4096,          // Maximum tokens to generate
  "stream": true,              // Enable streaming responses
  "n": 1,                      // Number of completions
  "stop": ["\\n", "END"],      // Stop sequences
  "logprobs": true             // Include log probabilities
}

Vision Capabilities

OpenAI's vision-capable models support image inputs:

{
  "messages": [
    {
      "role": "user",
      "content": [
        {"type": "text", "text": "What's in this image?"},
        {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,..."}}
      ]
    }
  ]
}

Function Calling

All GPT models support function calling, making them perfect for Claude Code:

{
  "model": "gpt-4o",
  "messages": [...],
  "tools": [{
    "type": "function",
    "function": {
      "name": "get_weather",
      "description": "Get the current weather",
      "parameters": {...}
    }
  }]
}

Troubleshooting

Rate Limits

OpenAI has different rate limits based on your usage tier:

• Monitor usage at platform.openai.com/usage
• Consider upgrading your plan for higher limits
• Implement exponential backoff for rate limit errors

API Errors

Common errors and solutions:

• 401 Unauthorized: Check your API key
• 429 Too Many Requests: You've hit rate limits
• 500 Server Error: OpenAI service issue, retry with backoff

Organization Access

If using an organization:

1. Ensure your API key has access to the organization
2. Add organization ID to your configuration
3. Check organization settings at platform.openai.com

Best Practices

1. Model Selection:

   • Use gpt-4o for most tasks (flagship model with vision support)
   • Use gpt-4.1 for coding-specific tasks (best instruction following)
   • Use gpt-4o-mini or gpt-4.1-nano for cost-efficient tasks
   • Use o3 or o3-pro for complex reasoning (no function calling)
   • Use o4-mini for fast reasoning tasks
   • Use gpt-3.5-turbo for simple, high-volume tasks

2. Cost Management:

   • Set usage limits in OpenAI dashboard
   • Monitor token usage in responses
   • Use smaller models when possible

3. Performance Optimization:

   • Implement caching for repeated queries
   • Use streaming for better perceived performance
   • Batch requests when possible

Pricing

Check openai.com/pricing for current pricing. As of July 2025:

• GPT-4o: Flagship model with best capabilities
• GPT-4o-mini: Cost-efficient version of GPT-4o
• GPT-4.1 family: Specialized coding models at various price points
• o3/o3-pro: Premium reasoning models (higher cost)
• o4-mini: Cost-efficient reasoning model
• GPT-3.5-turbo: Most affordable option for simple tasks

Pricing varies by model and usage tier. Enterprise customers should contact OpenAI for volume pricing.

Next Steps

• Quick Start Guide - Get running in 2 minutes
• Configuration Guide - Advanced configuration options
• Provider Comparison - Compare with other providers