What changed
Anthropic introduced Haiku 5.5 as its newest small model, aimed at high-volume, cost-sensitive workloads. The company highlights summaries, classification, browser use, and use alongside its larger models. These are vendor descriptions, not results from our lab.
The everyday takeaway
Try a short task you repeat often before assuming you need a larger model. Summaries and extraction are easier to audit when the original text is beside you.
What we have verified
The announcement describes the intended use. We have not run Haiku here or verified its performance claims.
A second perspective
Artificial Analysis notes strong small-model results alongside substantial token use at high effort. Read its pricing and evaluation caveats.