Anthropic has launched Claude Haiku 5.5, its latest small model designed for high-volume, cost-sensitive AI workloads.
The company describes it as its fastest and most capable Haiku model yet, while saying it costs significantly less to run than the previous-generation Haiku 4.5.
Haiku 5.5 is aimed at everyday tasks such as document summarisation, information classification, database queries, compaction and customer-support workflows.
These tasks can become expensive when businesses run them thousands or millions of times, making speed and low operating costs particularly important.
Built to Work Alongside Larger Claude Models
Anthropic has designed Haiku 5.5 to work alongside its larger Claude Opus 5.5 and Sonnet 5.5 models. In coding workflows, for example, a larger model can handle the main problem while Haiku 5.5 works as a subagent on smaller, repetitive tasks.
This approach lets developers reserve more powerful models for complex work while using Haiku 5.5 for tasks where speed and cost efficiency matter more.
First Haiku Model With Adjustable Effort
One of the biggest changes is the introduction of an adjustable effort setting. Haiku 5.5 is the first model in the Haiku family to offer this control.
Developers can choose how much computational effort the model should use. Lower effort can prioritise speed and cost, while higher effort allows the model to spend more resources on intelligence and reasoning.
Anthropic says Haiku 5.5 is 90% cheaper than Haiku 4.5 for requests up to 100,000 tokens and 50% cheaper for longer requests. The company notes that around 90% of requests to Haiku 4.5 fell within the 100,000-token range.
Improved Alignment and Safety
Anthropic also reports improvements in Haiku 5.5's alignment evaluations compared with Haiku 4.5, including fewer instances of behaviour the company classified as misaligned and a lower willingness to cooperate with misuse.
Availability
Claude Haiku 5.5 is now available through Anthropic's platform and major cloud services, including Amazon Web Services, Google Cloud, and Microsoft Azure. It is also available through Claude.ai and Claude Code.
The model is particularly positioned for businesses that need to run large numbers of relatively narrow AI tasks while keeping latency and costs under control.