Anthropic launches Claude Haiku 4.5, a smaller, cheaper AI model

Anthropic on Wednesday announced Claude Haiku 4.5, a small artificial intelligence model that’s available as a lower-cost offering for all of the company’s users. 

The model is fast and can outperform other larger models that were considered cutting edge just months ago, Anthropic said. 

Claude Haiku 4.5 is better at using computers than Claude Sonnet 4, for instance, which is a midsized model the company launched in May. It performs similarly to Claude Sonnet 4 and OpenAI’s most recent model, GPT-5, at coding, according to SWE-bench Verified, a test set that measures an AI system’s software coding abilities.

“It punches way above its weight,” Mike Krieger, Anthropic’s chief product officer, told CNBC in an interview. 

Claude Haiku 4.5 is available to Anthropic’s free users, and it’s now the cheapest model available to paid users.

Anthropic is an AI startup that develops a family of large language models called Claude. The company assigns new numbers to the models as they advance across generations, but the smallest model in the family is typically called Haiku, the midsized model is called Sonnet and the largest model is Opus. 

After OpenAI burst onto the scene with the launch of its chatbot ChatGPT in 2022, Anthropic launched a rival product, Claude, the following year. It’s powered by Anthropic’s family of models, and users can choose between free and paid tiers. 

The launch of Claude Haiku 4.5 comes just weeks after the company announced Claude Sonnet 4.5 in September and Claude Opus 4.1 in August. Anthropic is working to release another model, likely an updated version of Opus, by the end of this year or early next year, Krieger said.

For paid users, Haiku models are typically around one-third of the cost of Anthropic’s Sonnet models, while Sonnet models are one-fifth of the cost of its Opus models, Krieger said. Anthropic’s free users can still choose to use Claude Sonnet 4.5, but they’ll get more capacity out of Claude Haiku 4.5 since it’s smaller, he added. 

Claude Sonnet 4.5 is still Anthropic’s best-performing model, but the company said Claude Haiku 4.5 is ideal for users looking for fast, accurate answers.

“Even for my own use, even though it is not as smart as Sonnet, I’ve started defaulting to it on Claude, especially in the mobile app, because it’s just much faster getting an answer,” Krieger said. 

The two models can also work together. Anthropic said Claude Sonnet 4.5 can create multi-step plans to solve complex problems, and Claude Haiku 4.5 can complete subtasks within those plans, for example. 

Running the models in parallel could be particularly useful for businesses that want to use AI to tackle longer-term projects, Krieger said.

“You could have Haiku monitoring financial streams of data – and because it’s a smaller, cheaper, faster model, it can do that at a higher volume – and then pass off its early insights to Sonnet to do some deeper analysis,” he said. 

Anthropic, which was No. 4 on CNBC’s 2025 Disruptor 50 list and is valued at $183 billion, serves more than 300,000 business customers. Its annual revenue run rate is approaching $7 billion this month, according to an Anthropic spokesperson.

The company has been racing to keep up with competitors like Google and OpenAI, whose valuation has swelled to $500 billion. Following the launch of GPT-5 in August, OpenAI has inked several multibillion-dollar infrastructure deals and released a short-form video app called Sora.

The breakneck pace of the industry doesn’t afford Anthropic much time to get comfortable after a launch. While the company was carrying out the training for Claude Sonnet 4.5, it had already kicked off work on Claude Haiku 4.5.

“We’re really firing on all cylinders,” Krieger said. 

WATCH: Anthropic unveils overseas hiring push as rivalry with OpenAI goes global