Anthropic Haiku 5.5 cuts prices 10x and improves computer use
Haiku 5.5 drops input and output pricing to a tenth of the previous model and scores 72.4% on OSWorld computer use versus OpenAI's small model at 48.9%, positioned as a cheap subagent under Opus and Sonnet.
Anthropic just cut the price of its cheapest model 10x and the new one beats OpenAI's small model at operating a computer. 72.4% vs 48.9% on OSWorld. Run the actual numbers on the 10x. Haiku 4.5 charged $1 per million input tokens and $5 on output. Haiku 5.5 charges $0.10 and $0.50 for prompts under 100K tokens, which covers 90% of the traffic the old model served. Cache reads fell from a dime to a penny. So an always-on agent chewing through a million tokens an hour costs about $2.40 a day in input. A tireless digital worker, priced under your morning coffee. This is the part of the AI race that actually hits invoices. Frontier models win benchmarks and headlines. Small models win deployments, because most production AI is grunt work. Summaries, classification, routing. The question there was never "how smart" but "how cheap per task," and that number just dropped an order of magnitude. Haiku 4.5 shipped twelve months ago. Everything else you buy inflates 3% a year. Intelligence just deflated 90% in one. And the pairing pattern matters more than the price. Anthropic explicitly positions Haiku as the subagent under Opus and Sonnet. The expensive model plans, the cheap model executes a hundred tasks in parallel. That's an org chart, and the labor at the bottom of it just got 10x cheaper overnight.
This post quotes another post: x.com/1943306828697550848/status/210789403962627…