On Wednesday, Anthropic released Claude Haiku 4.5, the newest version of its smallest model, billed as offering similar performance to Sonnet 4 “at one-third the cost and more than twice the speed,” per a company blog post. While the
Anthropic sites a range of new benchmark results to back up those performance claims. In the company’s testing, Haiku scored 73% on SWE-Bench verified and 41% on the command-line-focused Terminal-Bench — below Sonnet 4.5, but on par with Sonnet 4, GPT-5 and Gemini 2.5 in each case. Tests show similar results on benchmarks for tool use, computer use, and visual reasoning.
The new version of Haiku will become the default for all free Anthropic plans and the company believes it will be particularly appealing for free versions of AI products, where it can provide significant capabilities while minimizing server loads. The lightweight nature of the model also means it’s easier to deploy multiple Haiku agents in parallel or in combination with a more sophisticated model.
In a statement to press, Anthropic CPO Mike Krieger said the Haiku would make new styles of deployment possible in production for the first time. “It’s is opening up entirely new categories of what’s possible with AI in production environments – with Sonnet handling complex planning while Haiku-powered sub-agents execute at speed,” Krieger said. “We’re giving people a complete agent toolbox where each model has the right combination of intelligence, speed, and cost for different parts of the job.”
The most immediate applications are likely to come in software development tools, where Claude Code is already commonly used and latency is often a critical factor. In statements provided by Anthropic, Zencoder CEO Andrew …