Anthropic has launched Claude Haiku 5.5, a new small-scale model designed for repetitive tasks like summarization and classification. The model features an adjustable effort setting and is available on the Claude Platform, as well as via Amazon Web Services, Google Cloud, and Microsoft Azure.
Pricing for Haiku 5.5 is set at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens. This represents a 90% reduction compared to the previous Haiku 4.5 model. Longer prompts receive a 50% discount.
The company also reduced the cost of cache reads for the midsize Claude Sonnet 5.5 model, cutting the price from $0.20 to $0.10 per million tokens. Anthropic estimates this change will lower the total cost of most agentic workloads on the model by approximately 20%.
Asana tested the new model for its AI Teammates agent, reporting a 30% reduction in task completion latency and up to 2.5 times faster inference per agent turn. In benchmark testing, Haiku 5.5 outperformed OpenAI’s GPT-6 Luna on several metrics, including OSWorld 2.1 and Terminal-Bench 4.0.
