Anthropic launches Claude Haiku 5.5, with average running costs reduced by about 75% compared to the previous generation

Odaily reports: Anthropic has announced the launch of Claude Haiku 5.5, positioned as its lowest-cost, fastest, and most capable small model to date, primarily targeting high-volume use cases such as summarization, context compression, database queries, classification, sub-agents, real-time customer service, and browser operations. Haiku 5.5 also becomes the first model in the Haiku series to support adjustable reasoning intensity.

In terms of pricing, Haiku 5.5's average running costs are about 75% lower than Haiku 4.5, with input and output prices within 100K tokens at $0.10 and $0.50 per million tokens, respectively. Additionally, Anthropic has reduced Sonnet 5.5's cache read pricing by 50%, which is expected to lower costs for most of its Agent tasks by about 20%; it will also provide Max 5x, Max 20x, and Team users with monthly API credits of $100, $200, and up to $500, respectively.