Anthropic has introduced Claude Haiku 5.5, positioning it as the company's fastest and most cost-effective small language model to date. The model is designed for high-volume tasks, including document summarization, database querying, and real-time customer support, as well as browser automation.
Pricing for developers 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% price reduction compared to the previous Haiku 4.5 model. Anthropic estimates an average cost saving of approximately 75% for users.
In performance benchmarks, Haiku 5.5 achieved a 72.4% success rate on OSWorld 2.1, outperforming OpenAI's GPT Luna, which scored 48.9%. On Terminal-Bench 4.0, the model reached a 39.2% success rate, compared to 16.4% for Luna and 0% for its predecessor.
The model also recorded a score of 1620 on the GDPval-AA v2.1 benchmark, surpassing the 1437 score of GPT Luna and the 735 score of Haiku 4.5. Additionally, Haiku 5.5 introduces an adjustable effort setting, allowing users to balance computational costs against model intelligence.
Anthropic also reduced the cache-read price for Sonnet 5.5 by 50%, setting it at $0.10 per million tokens. This release follows the launch of Opus 5.5 on September 22 and Sonnet 5.5 on September 28, completing the company's planned rollout of the Claude 5.5 series.
