Anthropic launches Cloud Haiku 4.5 model for smaller, faster artificial intelligence

Latest Claude Generative AI modelHaiku 4.5, Anthropic said in a press release on Wednesday, has the same encoding power as the company’s Sonnet 4 model in a smaller, faster package. The new form is now available to everyone, and will be the default form for free users Claude.i.

Anthropic says the Haiku 4.5 is much faster than the Sonnet 4, but at a third the cost. When in use Cloud for Chromean extension that gives Chrome users AI capabilities in their browser, Anthropic said Haiku 4.5 is faster and better at proxying tasks.

Don’t miss any of our unbiased technical content and lab reviews. Add CNET As Google’s preferred source.

Since Haiku 4.5 is a small model, it can be deployed as a subagent of Sonnet 4.5. So, while Sonnet 4.5 plans and organizes complex projects, smaller Haiku subagents can finish other tasks in the background. For programming tasks, Sonnet can handle high-level thinking while Haiku handles other tasks such as refactoring and migration. For financial analysis, Sonnet can perform predictive modeling while Haiku monitors data flows and tracks regulatory changes, market signals and portfolio risk. On the research side of things, Sonnet can handle comprehensive analysis while Haiku does literature review, data collection, and document compilation from multiple sources.

Haiku’s speed also helps on the chatbot side, as it handles requests faster.

“Haiku 4.5 is the latest iteration of our smallest model, designed for everyone who wants the superior intelligence, trustworthiness and creative partnership of Claude in a lightweight package,” Anthropic CEO Mike Krieger said in a statement provided to CNET.

Given the high cost of training and deploying AI models, companies have been looking for ways to roll out smaller, more efficient models that still perform well. An AI query consumes much more energy than a Google search, but it depends on the size of the AI ​​model. A large model with over 405 billion parameters can eat up 6,706 joules of energyThat’s enough to run a microwave for eight seconds, according to an MIT Technology Review report. However, a small model, with eight billion parameters, might only consume 114 joules of energy, which is like running a microwave for a tenth of a second. Google search can be used 1,080 joules of energy.

Allowing smaller, more efficient models to take on the burden of simpler queries or background tasks can result in significant server cost savings. ChatGPT-5, for example, You can move between modelsproviding instant answers to light questions to leverage more power for complex queries. Energy saving measures are essential as AI companies need to be able to catch up The potential trillions to be spent in data center investments.