Anthropic launches Claude Haiku 5.5, averaging 75% cheaper than Haiku 4.5

Anthropic launched Claude Haiku 5.5 on October 7, 2026, describing it as "the cheapest, fastest, and most capable small model we've ever released." The model costs around 75% less to run on average than Claude Haiku 4.5, dropping to 90% less on prompts up to 100K tokens. It introduces the first effort setting for the Haiku line, letting users trade cost for accuracy. Anthropic still recommends Sonnet 5.5 and Opus 5.5 for complex agentic coding tasks such as Terminal-Bench 4.0. Benchmark results show large jumps: Haiku 5.5 scored 72.4% on OSWorld 2.1, a computer-use agent benchmark, up from 15.7% for Haiku 4.5, and 46.4% on Chartography, a visual chart-reading benchmark, up from 6.4%. Asana reported over 30% lower task-completion latency and up to 2.5x faster inference per agent turn. Cursor added the model to its IDE on launch day, noting it costs 10x less than Haiku 4.5 on shorter requests and can be enabled via Cursor Settings under Models.