Anthropic Cuts Claude AI Costs by 90% With New Prompt Caching
Anthropic introduces a new Prompt Caching feature for Claude that cuts costs by up to 90% and speeds up response times by 50%. The beta tool helps developers save money and time on complex, repetitive tasks.
Anthropic introduces a new Prompt Caching feature for Claude that significantly reduces costs and speeds up processing for complex tasks. Available now in public beta through the Anthropic API, this tool allows the AI to remember and reuse large amounts of contextual data from previous interactions instead of reprocessing the same information.
The new caching capability slashes usage costs by up to 90 percent and doubles processing speed by cutting latency in half. This provides a major advantage for businesses of all sizes that rely on repetitive tasks involving extensive background information, detailed instruction sets, or large document processing.
Developers access this feature immediately with Claude 3.5 Sonnet and Claude 3 Haiku models. Anthropic highlights several ideal use cases for the technology, including long conversational agents, coding assistants that maintain a summarized codebase, and agentic tool use that requires multiple iterative API calls.