Anthropic Silently Dropped Prompt Cache TTL from 1 Hour to 5 Minutes

📰 Dev.to AI

Anthropic reduced prompt cache TTL from 1 hour to 5 minutes, impacting Claude API cache hit rates

intermediate Published 18 Apr 2026

Action Steps

Check your Claude API code for prompt caching configuration
Update the cache TTL to 5 minutes or a suitable value for your application
Test your API calls to ensure the new cache TTL is being respected
Monitor your cache hit rate to optimize performance
Consider implementing a cache warming strategy to mitigate the impact of the reduced TTL

Who Needs to Know This

Developers using Claude API with prompt caching need to update their code to account for the changed TTL, to avoid decreased cache hit rates and potential performance issues

Key Insight

💡 Anthropic's reduction of prompt cache TTL from 1 hour to 5 minutes requires developers to update their Claude API code to maintain optimal performance