OpenAI Compatibility

Why OpenAI Compatibility?

The OpenAI Chat Completions API has become the de facto standard for AI applications. By implementing this same interface, Vizra ADK instantly becomes compatible with thousands of existing tools, libraries, and workflows without any code changes.

Existing Tools

Use with LangChain, LlamaIndex, Vercel AI SDK, and countless other libraries

Client Apps

Works with ChatGPT clients, mobile apps, browser extensions, and desktop tools

Zero Migration

Just change the base URL - everything else works exactly the same

API Endpoint

OpenAI Compatible Endpoint

POST /api/vizra-adk/chat/completions

This endpoint accepts the exact same request format as OpenAI’s Chat Completions API.

Quick Start

Ready to try it? Here are examples in different languages:

cURL
Laravel Http
JavaScript
Python

Terminal

curl -X POST http://your-app.com/api/vizra-adk/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "your-agent-name",
    "messages": [
      {"role": "user", "content": "Hello! Tell me about yourself."}
    ],
    "temperature": 0.7,
    "max_tokens": 500
  }'

use Illuminate\Support\Facades\Http;

$response = Http::post('http://your-app.com/api/vizra-adk/chat/completions', [
    'model' => 'your-agent-name',
    'messages' => [
        ['role' => 'user', 'content' => 'Hello! Tell me about yourself.']
    ],
    'temperature' => 0.7,
    'max_tokens' => 500
]);

$result = $response->json();
echo $result['choices'][0]['message']['content'];

const response = await fetch('http://your-app.com/api/vizra-adk/chat/completions', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    model: 'your-agent-name',
    messages: [
      { role: 'user', content: 'Hello! Tell me about yourself.' }
    ],
    temperature: 0.7,
    max_tokens: 500
  })
});

const data = await response.json();
console.log(data.choices[0].message.content);

import requests

response = requests.post(
    'http://your-app.com/api/vizra-adk/chat/completions',
    json={
        'model': 'your-agent-name',
        'messages': [
            {'role': 'user', 'content': 'Hello! Tell me about yourself.'}
        ],
        'temperature': 0.7,
        'max_tokens': 500
    }
)

result = response.json()
print(result['choices'][0]['message']['content'])

Using with Existing Libraries

OpenAI SDK (Python)
LangChain

from openai import OpenAI

# Just change the base_url to point to your Vizra ADK instance
client = OpenAI(
    api_key="not-needed",  # Vizra ADK doesn't require API keys
    base_url="http://your-app.com/api/vizra-adk"
)

response = client.chat.completions.create(
    model="your-agent-name",
    messages=[
        {"role": "user", "content": "What can you help me with?"}
    ]
)

print(response.choices[0].message.content)

from langchain_openai import ChatOpenAI

# Use your Vizra agents with LangChain
llm = ChatOpenAI(
    model="your-agent-name",
    openai_api_key="not-needed",
    openai_api_base="http://your-app.com/api/vizra-adk",
    temperature=0.7
)

response = llm.invoke("What's the weather like today?")
print(response.content)

Configuration

Configure model-to-agent mapping to make your agents accessible via familiar OpenAI model names:

config/vizra-adk.php

return [
    // ... other config

    /**
     * OpenAI API Compatibility Configuration
     * Maps OpenAI model names to your agent names
     */
    'openai_model_mapping' => [
        // Default mappings for OpenAI models
        'gpt-4' => env('VIZRA_ADK_OPENAI_GPT4_AGENT', 'chat_agent'),
        'gpt-4-turbo' => env('VIZRA_ADK_OPENAI_GPT4_TURBO_AGENT', 'chat_agent'),
        'gpt-3.5-turbo' => env('VIZRA_ADK_OPENAI_GPT35_AGENT', 'chat_agent'),
        'gpt-4o' => env('VIZRA_ADK_OPENAI_GPT4O_AGENT', 'chat_agent'),
        'gpt-4o-mini' => env('VIZRA_ADK_OPENAI_GPT4O_MINI_AGENT', 'chat_agent'),

        // Add your own custom mappings here
        // 'my-custom-model' => 'my_specialized_agent',
        // 'claude-3-opus' => 'advanced_reasoning_agent',
        // 'gpt-4-vision' => 'image_analysis_agent',
    ],

    /**
     * Default agent when no mapping is found
     * Used for unmapped OpenAI models (gpt-*)
     */
    'default_chat_agent' => env('VIZRA_ADK_DEFAULT_CHAT_AGENT', 'chat_agent'),
];

How Model Resolution Works:

First checks for exact match in openai_model_mapping
If model starts with gpt-, uses default_chat_agent
Otherwise, treats the model name as the agent name directly

This means you can use model: "your_agent_name" directly without any mapping.

You can customize mappings via environment variables or by publishing the config file with php artisan vendor:publish --tag=vizra-adk-config.

Streaming Support

Enable real-time streaming responses by setting "stream": true in your request:

JavaScript
Python

async function streamResponse() {
  const response = await fetch('/api/vizra-adk/chat/completions', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      model: 'my-agent',
      messages: [{ role: 'user', content: 'Tell me a long story' }],
      stream: true,
      temperature: 0.8
    })
  });

  const reader = response.body.getReader();
  const decoder = new TextDecoder();

  while (true) {
    const { done, value } = await reader.read();
    if (done) break;

    const chunk = decoder.decode(value);
    const lines = chunk.split('\n');

    for (const line of lines) {
      if (line.startsWith('data: ')) {
        const data = line.slice(6);
        if (data === '[DONE]') return;

        try {
          const parsed = JSON.parse(data);
          const content = parsed.choices[0]?.delta?.content;
          if (content) {
            process.stdout.write(content); // Stream to console
            // Or update your UI in real-time
          }
        } catch (e) {
          // Handle parsing errors
        }
      }
    }
  }
}

streamResponse();

import requests
import json

def stream_completion():
    response = requests.post(
        'http://your-app.com/api/vizra-adk/chat/completions',
        json={
            'model': 'my-agent',
            'messages': [{'role': 'user', 'content': 'Write a poem'}],
            'stream': True,
            'temperature': 0.7
        },
        stream=True
    )

    for line in response.iter_lines():
        if line.startswith(b'data: '):
            data = line[6:].decode('utf-8')
            if data == '[DONE]':
                break

            try:
                chunk = json.loads(data)
                content = chunk['choices'][0]['delta'].get('content')
                if content:
                    print(content, end='', flush=True)
            except json.JSONDecodeError:
                continue

stream_completion()

Supported Parameters

The OpenAI compatibility layer supports all major ChatGPT parameters:

Parameter	Description
`model`	Agent name or mapped model name
`messages`	Array of conversation messages
`stream`	Enable streaming responses
`temperature`	Creativity level (0.0 - 2.0)
`max_tokens`	Maximum response length
`top_p`	Nucleus sampling parameter
`user`	User identifier for sessions

Response Format

Responses match OpenAI’s format exactly, ensuring perfect compatibility:

Standard Response

Non-streaming Response

{
  "id": "chatcmpl-AbCdEfGhIjKlMnOpQrStUvWxYz",
  "object": "chat.completion",
  "created": 1677858242,
  "model": "your-agent-name",
  "system_fingerprint": null,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! I'm your Vizra agent, ready to help!"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 15,
    "total_tokens": 27
  }
}

Streaming Response

Server-Sent Events Format

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677858242,"model":"your-agent","choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677858242,"model":"your-agent","choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677858242,"model":"your-agent","choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}]}

data: {"id":"chatcmpl-123","object":"chat.completion.chunk","created":1677858242,"model":"your-agent","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Error Handling

Error responses also match OpenAI’s format for seamless compatibility:

Error Response Format

{
  "error": {
    "message": "The model 'unknown-agent' does not exist or you do not have access to it.",
    "type": "not_found_error",
    "code": "model_not_found"
  }
}

Status Code	Description
400 - Bad Request	Invalid request format or missing required fields
404 - Not Found	Agent/model not found or not registered
500 - Server Error	Internal error during agent execution

Tips & Best Practices

Agent Naming Strategy

Map commonly used OpenAI model names to your best agents to make migration seamless. For example, map gpt-4 to your most advanced agent.

Performance Optimization

Use the user parameter to maintain persistent sessions and memory across conversations for more personalized responses.

Development Workflow

Test your OpenAI compatibility with existing tools during development. Most AI applications allow changing the base URL for easy integration testing.

Direct Agent Access

You can use model: "your_agent_name" directly without any mapping configuration.

Getting Started

Core Concepts

Agents

Tools

Advanced Features

Integrations

Testing & Observability

API Reference

OpenAI Compatibility

Why OpenAI Compatibility?

Existing Tools

Client Apps

Zero Migration

API Endpoint

Quick Start

Using with Existing Libraries

Configuration

Streaming Support

Supported Parameters

Response Format

Standard Response

Streaming Response

Error Handling

Tips & Best Practices

Agent Naming Strategy

Performance Optimization

Development Workflow

Direct Agent Access

Next Steps

Get Started

Learn About Agents

Getting Started

Core Concepts

Agents

Tools

Advanced Features

Integrations

Testing & Observability

API Reference

​Why OpenAI Compatibility?

Existing Tools

Client Apps

Zero Migration

​API Endpoint

​Quick Start

​Using with Existing Libraries

​Configuration

​Streaming Support

​Supported Parameters

​Response Format

​Standard Response

​Streaming Response

​Error Handling

​Tips & Best Practices

Agent Naming Strategy

Performance Optimization

Development Workflow

Direct Agent Access

​Next Steps

Get Started

Learn About Agents

Why OpenAI Compatibility?

API Endpoint

Quick Start

Using with Existing Libraries

Configuration

Streaming Support

Supported Parameters

Response Format

Standard Response

Streaming Response

Error Handling

Tips & Best Practices

Next Steps