Anthropic has launched Claude Haiku 5.5, the third model in its Claude 5.5 family and the company's fastest model at standard speed. It is aimed at high-volume and latency-sensitive work such as summarization, classification, database queries, browser use, customer support and subagent tasks.
Pricing depends on prompt length. For prompts up to 100,000 tokens, Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens. Above 100,000 tokens, pricing rises to $0.50 input and $2.50 output. Anthropic says average running cost is around 75% lower than Haiku 4.5 after accounting for prompt distribution and tokenizer differences. It also cut Sonnet 5.5 cache-read pricing by half.
Haiku 5.5 is the first Haiku model with an adjustable effort setting. Anthropic reports large benchmark gains over Haiku 4.5, including 72.4% on the OSWorld 2.1 offline subset and 39.2% on Terminal-Bench 4.0. These benchmark figures are primarily vendor-reported and should be distinguished from independent production validation, although Reuters independently confirmed the model launch.
The model is available through the Claude Platform as claude-haiku-5-5 and through AWS, Google Cloud and Microsoft Azure. Anthropic is also adding computer-use and browser-use beta support to its Python and TypeScript SDKs. The release makes the small-model tier more capable and substantially cheaper, which may make multi-agent architectures more economical when many routine subagent calls do not require a larger model.