Anthropic ships Claude Haiku 5.5: about 75% cheaper on average than Haiku 4.5

Small model aimed at high-volume work (summaries, classification, sub-agents); Anthropic also cut Sonnet 5.5 cache-read prices by 50%. AlphaSense reported a gain on its Ask-in-Document eval (0.84 vs 0.76).

Why it matters for deal teams: The economics of 'read every amendment in the book every night' just changed. Cheap sub-agents for extraction, a bigger model for judgment.