Photo Credit: Getty Images
Anthropic has officially launched Claude Haiku 4.5, the newest version of its lightweight AI model designed to deliver high performance at a fraction of the cost. The model, which becomes the new default for free users, is touted as both faster and more affordable, ideal for those seeking quick, accurate responses without sacrificing reliability.
According to Anthropic, Haiku 4.5 matches or even surpasses the performance of older midsized models. It performs competitively with Claude Sonnet 4 and OpenAI's GPT-5 in software development tasks, scoring 73% on SWE-Bench Verified, a benchmark for coding ability. "It punches way above its weight," said Mike Krieger, Anthropic's Chief Product Officer.
Haiku 4.5's biggest edge is speed. At more than twice the processing pace of Sonnet 4.5 and roughly one-third the cost, the model opens new possibilities for both consumers and enterprise users. It can efficiently handle subtasks in tandem with larger models, making it ideal for AI workflows that balance cost, intelligence, and scale.
Anthropic's system of models, Haiku, Sonnet, and Opus, caters to different performance needs, with Haiku being the smallest, fastest, and cheapest. The company has been on a rapid release cycle this year, unveiling Sonnet 4.5 in September and Opus 4.1 in August. An upgraded Opus is also expected later this year.
The San Francisco-based firm, valued at $183 billion, now serves over 300,000 business clients, with revenue projected to hit $7 billion annually. This growth positions Anthropic as a fierce rival to OpenAI and Google DeepMind, both racing to dominate the AI industry.
Krieger said the smaller model's low latency and high scalability make it perfect for real-time applications such as financial monitoring, software debugging, and customer support. "You could have Haiku scanning large data streams while Sonnet performs deeper analysis," he explained.
As Anthropic expands globally and strengthens its model lineup, Claude Haiku 4.5 signals the company's strategy to democratize access to advanced AI, bringing enterprise-grade intelligence to users at unprecedented speed and affordability.
