Prompt cache configuration

To enable prompt caching:

  • For v1/messages and v1/messages/streaming, add a cache control object after each content object with these properties:
    • "type": "cache_control"
    • "data": true
  • For v2/messages, add a cache control object after each content object with these properties:
    • "cachePoint":{"type":"default"}"

The cache block stays the same for all requests during the five-minute TTL (Time To Live).

Prompt cache for v2/messages and v2/messages/streaming

o	{
  "modelId": "claude-sonnet-4-20250514-v1:0",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "text": "ITEM: FILTER-HYD-P3 | Hydraulic Filter P3 | $34.90 | Stock: 87\nITEM: PUMP-CENT-2HP | Centrifugal Pump 2HP | $1240.00 | Stock: 12\n... [5,000+ items] ..."
        },
        {
          "cachePoint": {
            "type": "default"
          }
        },
        {
          "text": "What is the stock level and price for the Hydraulic Filter P3 Series?"
        }
      ]
    }
  ]
}