tl;dr: Anthropic released Claude Haiku 5.5, its fastest and cheapest small model, priced ~75% lower than Haiku 4.5 ($0.10/$0.50 per million input/output tokens for prompts under 100k), with an adjustable effort setting and benchmarks competitive with GPT-6 Luna on tasks like OSWorld and Terminal-Bench. Alongside the launch, Anthropic halved Sonnet 5.5 cache-read prices (cutting agentic workload costs ~20%) and added monthly API credits ($100–$500) for Max and Team subscribers. SDKs now support computer and browser use in beta.