Anthropic Debuts Claude Haiku 5.5, Cutting Workload Operating Costs by 75%

Anthropic officially introduced Claude Haiku 5.5, positioning it as the company's fastest and most capable small-footprint model to date, designed specifically for high-volume enterprise workloads, subagent orchestration, and real-time tasks.
According to Anthropic, typical enterprise workloads cost roughly 75% less to run compared to predecessor Haiku 4.5, driven by significant token efficiency gains alongside aggressive API price reductions starting at $0.10 per million input tokens and $0.50 per million output tokens for requests under 100,000 tokens.
The release marks a targeted push into cost-sensitive production applications—such as document extraction, live agent routing, and automated classification—intensifying price competition against rival low-latency models while expanding deployment availability across Anthropic's API and Amazon Web Services ($AMZN) Bedrock.