Claude Haiku 5.5: a faster model for short, repetitive tasks
Anthropic introduced Haiku 5.5 for high-volume work such as summaries, classification and subagents. Anthropic reports an average run-cost reduction of around 75% compared with Haiku 4.5.
Anthropic introduced Claude Haiku 5.5 on October 7, 2026. It is designed for large volumes of relatively short tasks, including summaries, lookups, database queries and classification. Anthropic also positions it as a fast subagent alongside Opus 5.5 and Sonnet 5.5.
The company reports that Haiku 5.5 costs around 75% less to run on average than Haiku 4.5. That is a model execution-cost comparison, not a 75% reduction in the price of a Claude subscription. Actual costs depend on the task, token use and how you access the model.
Where it can help: let Haiku handle repetitive first-pass work—finding relevant passages, sorting incoming requests or drafting short summaries—while a stronger model or a person reviews important conclusions.
Our advice: test the model on representative tasks from your own workflow. Measure not only speed and cost, but also errors and the time needed to correct the output.
Key takeaways
- Built for high-volume, cost-sensitive work such as summaries and classification
- Anthropic reports around 75% lower average run costs than Haiku 4.5
- Can work as a fast subagent alongside Opus 5.5 and Sonnet 5.5
- Lower model execution costs do not automatically mean lower subscription fees