Anthropic has released Claude Haiku 5.5, its fastest and most affordable small model to date, featuring major benchmark gains and token price reductions of up to 90 percent. Built for high-volume and cost-sensitive workloads like data queries, classification, and customer support, the new model is available now across major cloud platforms including Amazon Web Services, Google Cloud, and Microsoft Azure.
Benchmarks show a substantial performance leap over the previous generation, with Haiku 5.5 scoring 1,620 on the GDPval-AA v2.1 knowledge benchmark and achieving 72.4 percent on the OSWorld-2.1 computer use test. Additionally, the model introduces adjustable reasoning levels, allowing users to balance speed and cost, while outperforming OpenAI's budget model GPT-6 Luna across tested categories.
Alongside the Haiku release, Anthropic is cutting cache read costs for Sonnet 5.5 by 50 percent to $0.10 per million tokens and rolling out monthly API credits ranging from $100 to $500 for subscribers. These concurrent pricing updates demonstrate that the competitive AI pricing arms race among leading labs is intensifying.