Prompt Caching Fails to Reduce Quota in Claude Code Pro Max

Claude Code Pro Max 5x users are experiencing rapid quota exhaustion because cached tokens count at full rate against rate limits, eliminating the intended cost-benefit of prompt caching. This fundamental mismatch between API implementation and documented behavior undermines the viability of long-context sessions. The issue demands urgent clarification on how cache_read tokens are actually accounted for in the quota system.