Anthropic’s smallest model jumped a generation and a price tier
Anthropic released Claude Haiku 5.5 on Wednesday, calling it its cheapest, fastest and most capable small model, and priced it far below the one it replaces. For prompts up to 100,000 tokens — around nine in ten requests to the old model — it costs 90% less than Haiku 4.5: $0.10 per million input tokens and $0.50 per million output, with cache reads at $0.01. Above that threshold the discount is 50%. Anthropic says a typical Haiku workload now runs about 75% cheaper, and part of the gap is a new tokenizer that uses slightly more tokens per task.
The capability jump is not marginal. On Humanity’s Last Exam it scores 45.9% without tools against Haiku 4.5’s 10.2%, and 57.4% with them; on OSWorld 2.1, which measures an agent operating a real computer, it reaches 72.4% against 15.7%; and on Terminal-Bench 4.0 it goes from 0.0% to 39.2%. It is the first Haiku-class model with an adjustable effort setting. Anthropic also halved the price of Sonnet 5.5 cache reads, cutting its cost on most agentic work by around 20%, and added monthly API credits for Max and Team subscribers.
WHY IT MATTERS
The floor of the model market just dropped: a small model that clears benchmarks Haiku 4.5 failed outright, at a tenth of the cost, which changes what is worth routing to a frontier model at all. The OSWorld and Terminal-Bench scores matter most for anyone building agents — a cheap model that can drive a browser and a shell is the difference between an agent you demo and one you leave running. Anthropic’s own framing points the same way: it sells Haiku 5.5 as a subagent and compactor for Opus rather than a rival to it, so the pattern that pays is the hybrid one the decider-model releases are pushing from the other side. The caveat deserves one line: the numbers are Anthropic’s own, and the higher Sonnet and Opus scores on the same agentic tasks are the ones it wants you to buy.