Anthropic launches Claude Haiku 5.5, the first Haiku model with effort controls, for high-volume, cost-sensitive tasks like summaries and classification (Anthropic)

Anthropic has unveiled Claude Haiku 5.5, their latest small model with adjustable effort settings. This model is tailored for high-volume, cost-sensitive tasks like summaries, compactions, database queries, and classification requests. Haiku 5.5 works well alongside Opus 5.5 and Sonnet 5.5 as a subagent for coding tasks, and its speed makes it ideal for speed-sensitive applications such as live customer support and browser use.

Priced at about 75% less than its predecessor Haiku 4.5, Haiku 5.5 offers significant cost savings while still delivering impressive performance. Along with this launch, Anthropic is also reducing the price of Claude Sonnet 5.5's cache reads by half, making it around 20% cheaper for most agentic work. Additionally, a new monthly API credit is being introduced for Claude Max and Team subscribers to support users in building new agents and applications using the Claude Platform.

Claude Haiku 5.5 shows major improvements in alignment evaluations compared to Haiku 4.5, with fewer instances of misaligned behavior and a lower willingness to cooperate with misuse. Its cybersecurity safeguards are more restrictive than those of Haiku 4.5 but less so than those of other recent models. Haiku 5.5's biology safeguards are consistent with those of Sonnet 5.5, Sonnet 5.5, and Opus 5.5, allowing research biology questions while restricting potentially harmful requests.

Available on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure, Haiku 5.5 can be accessed through the Claude Platform using the claude-haiku-5-5. While larger models like Sonnet 5.5 and Opus 5.5 are better suited for complex agentic coding tasks, Haiku 5.5 excels in narrowly scoped tasks such as compaction, summarization, or subagent work.