Anthropic launches Claude Haiku 5.5 as its fastest, cheapest small model, with lower Sonnet 5.5 agent costs

Anthropic introduced Claude Haiku 5.5, positioning it as its lowest-cost and fastest small model for high-volume tasks like summaries, classification, database queries, customer support, and browser use. The company says Haiku 5.5 costs about 75% less to run than Haiku 4.5, while Sonnet 5.5 cache reads are being cut in half to reduce agentic workload costs.

The launch reflects a broader AI market shift toward model tiering, where cheaper small models handle routine work while larger models focus on complex reasoning. Lower API and caching costs could make multi-agent systems more practical for businesses, especially in coding, support, and workflow automation.