Batch API

📖 What is Batch API?

The Batch API allows you to process multiple AI requests in a single batch operation, which is cost-effective and efficient for large-volume processing. Instead of making individual API calls, you submit multiple messages at once, and the system processes them asynchronously. Once completed, the results are sent to your specified webhook URL.

Best suited for non-time-sensitive, high-volume operations.

Perfect for:

• Bulk content generation

• Data processing

• Translations

• Summarizations

• Any large-scale AI workflows

🔗 API Endpoint

POST <https://api.gtwy.ai/api/v2/model/batch/chat/completion>

📋 Request Parameters

Required Headers

Request Body

⚠️ Important:

If batch_variables is provided, its length must exactly match the batch array length.

💻 Example Request

curl --location '<https://api.gtwy.ai/api/v2/model/batch/chat/completion>' \\
--header 'pauthkey': 'YOUR_PAUTH_KEY' \\
--header 'content-type: application/json' \\
--data '{
  "webhook": {
    "url": "<https://your-webhook-url.com/endpoint>",
    "headers": {}
  },
  "batch": [
    "hi my name is husain",
    "hi my name is harsh"
  ],
  "bridge_id": "68b6ef6f4623c70889b02ebd",
  "batch_variables": [
    {
      "name": "husain"
    },
    {
      "name": "harsh"
    }
  ]
}'

💡 Tip: Always store batch_id and custom_id for tracking and debugging.

📬 Webhook Response

Webhook is triggered once after all batch items are processed (within 24 hours).

Example Webhook Payload

[
  {
    "data": {
      "id": "chatcmpl-CnKGinzA17Zo15BvFqMLP9WvydoGp",
      "content": "Hi Husain! How can I assist you today?",
      "model": "gpt-4o-mini-2024-07-18",
      "role": "assistant",
      "finish_reason": "completed"
    },
    "usage": {
      "input_tokens": 35,
      "output_tokens": 9,
      "total_tokens": 44
    },
    "batch_id": "batch_694112f91a7c8190b388404757b50f40",
    "custom_id": "a3d8f7b2-4c1e-4a9f-9b5d-7e2c8f1a4d6b",
    "variables": {
      "name": "husain"
    }
  }
]

🎯 Using batch_variables

Attach metadata to each request for accurate response mapping.

Rules

1. Must match batch length

2. Index-based mapping

3. Any valid JSON

4. Returned unchanged in webhook