Anthropic launches Claude Haiku 5.5, the cheapest and fastest small-model AI

Anthropic released Claude Haiku 5.5 on Wednesday, describing it as the cheapest, fastest and most capable small model in its portfolio Decrypt. Developers pay $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, matching OpenAI’s rate for GPT-6 Luna and representing a 90% reduction from Haiku 4.5’s $1 / $5 pricing.

The model is aimed at high-volume, low-complexity workloads such as document summarisation, database queries, live customer-support chats and web-browser assistance. In benchmark tests Haiku 5.5 achieved 72.4% success on the OSWorld 2.1 offline subset and 39.2% on Terminal-Bench 4.0, both ahead of GPT-6 Luna’s 48.9% and 16.4% respectively, though Sonnet 5.5 scored 70.6% on the coding test. On the GDPval-AA v2.1 professional-work rating, Haiku 5.5 scored 1620, compared with Luna’s 1437 and Haiku 4.5’s 735.

Haiku 5.5 introduces an adjustable effort setting that lets users trade cost for smarter answers. Anthropic also halved Sonnet 5.5’s cache-read price to $0.10 per million tokens and is adding monthly API credits for Max and Team subscribers this week. The model is available now via the Claude website, Amazon Web Services, Google Cloud and Microsoft Azure under the name claude-haiku-5-5.

Our read: The pricing shift makes Haiku 5.5 the most economical small model on the market, while benchmark gains suggest it is competitive with OpenAI’s offering despite remaining behind Anthropic’s own Sonnet 5.5 on certain tasks.