Anthropic Launches Claude Haiku 5.5 With 75% Lower Pricing

Anthropic has officially launched Claude Haiku 5.5, positioning the new release as its most affordable, rapid, and capable small-language model to date. Designed primarily for high-volume, cost-sensitive workflows, the model targets repetitive tasks such as data compaction, summarization, database querying, and live customer support. Its architecture is optimized to operate efficiently as a subagent alongside Claude Opus 5.5 and Sonnet 5.5, particularly within complex coding agentic pipelines. The model introduces significant commercial and performance advancements. Anthropic reports that Haiku 5.5 reduces operational costs by approximately seventy-five percent compared to its predecessor, Haiku 4.5, while maintaining robust benchmark scores across knowledge work, multidisciplinary reasoning, and visual analysis. Notably, it is the first model in the Haiku series to feature an adjustable effort setting, allowing developers to dynamically balance computational cost against reasoning depth. Early adoption metrics align with these benchmarks, highlighting strong reliability for narrowly scoped, high-throughput applications. Alongside the model release, Anthropic is restructuring its pricing strategy to enhance platform value. Cache read costs for Claude Sonnet 5.5 have been halved to $0.10 per million tokens, effectively reducing overall agentic workload expenses by roughly twenty percent. These adjustments are complemented by a new tiered API credit allocation for Max and Team subscribers, granting up to $500 in pooled monthly credits to facilitate experimentation with custom agents and integrations. Safety and compliance frameworks have been simultaneously upgraded. Haiku 5.5 demonstrates marked improvements in alignment evaluations, significantly reducing instances of misaligned outputs and misuse cooperation. Its cybersecurity safeguards balance operational flexibility with risk mitigation, permitting broader defensive research while maintaining strict blocks against offensive penetration testing. Biological and cybersecurity verification protocols remain aligned with those applied to the Opus and Sonnet series, with enterprise access governed by Anthropic's dedicated verification programs. Claude Haiku 5.5 is immediately available across major cloud providers, including Amazon Web Services, Google Cloud, and Microsoft Azure, as well as the native Claude Platform. To accelerate developer adoption, Anthropic has updated its Python and TypeScript software development kits to support beta computer and browser automation features, which pair seamlessly with the model's speed and pricing advantages. The launch solidifies Anthropic's strategy to capture the enterprise agentic computing market by delivering scalable, economically viable infrastructure for next-generation AI workflows.