"Performance That Exceeds Its Size Class"
Closing in on Industry Leader OpenAI
Artificial intelligence (AI) startup Anthropic has ignited the competition to attract subscribers by launching a new, cost-effective AI model.
On October 15 (local time), Anthropic announced the release of its latest compact AI model, Claude Haiku 4.5, stating, "A model that was recently at the cutting edge is now cheaper and faster."
Anthropic explained that compared to the Claude Sonnet 4 model released five months ago, the newly launched Claude Haiku 4.5 offers similar coding performance at just one-third of the cost and more than twice the speed.
The company also noted that Claude Haiku 4.5 even outperforms Sonnet 4 in certain tasks, such as computer applications.
In the SWE-bench Verified test, which evaluates AI models, Haiku 4.5 achieved a software engineering performance score of 73.3%, slightly higher than Sonnet 4 (72.7%) and OpenAI's latest model, GPT-5 (72.8%).
Mike Krieger, Anthropic's Chief Product Officer (CPO), said in an interview with CNBC on the same day, "Claude Haiku 4.5 delivers performance that far exceeds its class."
This model is available to free users and is offered to paid users at the lowest price.
Anthropic has been releasing large language model product lines under the Claude series, typically naming the smallest model Haiku, the mid-sized model Sonnet, and the largest model Opus.
On September 29, the company also released Claude Sonnet 4.5, the latest version of its mid-sized AI model. By introducing another new model in just about two weeks, Anthropic is closing in on industry leader OpenAI.
CPO Krieger emphasized that while Claude Sonnet 4.5 remains Anthropic's highest-performing model, the newly released Claude Haiku 4.5 is the ideal tool for users seeking faster and more accurate responses.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
