Key facts
- Anthropic launched Claude Haiku 5.5, its third AI model in a month.
- Haiku 5.5 is priced at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens.
- The new pricing represents a 90% reduction compared to Haiku 4.5's rates.
- Haiku 5.5 scored 72.4% on the OSWorld 2.1 test and 39.2% on Terminal-Bench 4.0.
- Anthropic also reduced the price for Sonnet 5.5's cache-read to $0.10 per million tokens.
- Anthropic has introduced Chinese-language options for its Claude chatbot.
Anthropic released Claude Haiku 5.5 on Wednesday, its third AI model in a month, ahead of a planned initial public offering (IPO). The company described Haiku 5.5 as its cheapest, fastest and most capable small model to date, designed for high-volume tasks like summarizing documents and querying databases, as well as speed-sensitive applications such as live customer support.
Developers will pay $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, a significant price reduction from Haiku 4.5's rates of $1 and $5. This represents a 90% decrease for Haiku 4.5's original rates, with prompts over 100,000 tokens receiving a 50% cut. Anthropic estimates the average saving for users will be around 75%.
Benchmarks indicate Haiku 5.5's capabilities, scoring 72.4% on the OSWorld 2.1 test for operating a computer and 39.2% on Terminal-Bench 4.0 for completing professional tasks. This positions it ahead of OpenAI's GPT-6 Luna on both tests, though behind Anthropic's own Sonnet 5.5 on the coding benchmark. Haiku 5.5 also features an adjustable effort setting, allowing users to balance cost with answer quality.
In addition to the new Haiku model, Anthropic has halved the cache-read price for Sonnet 5.5 to $0.10 per million tokens and is introducing monthly API credits for Max and Team subscribers. The company has also introduced Chinese-language options for its Claude chatbot in supported markets, though it continues to block users from mainland China.
