From the source
Anthropic released Claude Haiku 5.5, its fastest and most affordable small model, with benchmark scores far exceeding Haiku 4.5 and outperforming OpenAI's GPT-6 Luna across tested categories.
The model costs about 75 percent less than its predecessor on average, with up to 90 percent lower token prices for short prompts.
Anthropic also cut Sonnet 5.5 cache read costs by 50 percent and introduced monthly API credits for subscribers.


