Investing.com -- Amazon has restructured part of its partnership with artificial intelligence startup Anthropic, shifting from paying for its AI models based on computing hours to a token-based pricing model beginning next year, according to a report by The Information.

The report, citing people familiar with the matter, said the change could increase Amazon's costs for using Anthropic's large language models, including Claude. Amazon disputed that claim.

"Amazon and Anthropic share a multifaceted partnership grounded in technical collaboration, and we continue to foster that relationship and deepen our work together," an Amazon spokesperson said, adding that it was "incorrect that changes from our expanded collaboration will increase our costs."

Amazon uses Anthropic's models to power several AI offerings, including its Alexa shopping assistant, the Kiro coding tool and Quick, a workplace assistant.

The companies have been partners since 2023, when Amazon invested $4 billion in Anthropic in exchange for the startup making Amazon Web Services (AWS) its primary cloud provider and using Amazon's custom AI chips for training and inference. Earlier this year, Amazon agreed to invest up to an additional $25 billion in Anthropic.

Amazon has also expanded its AI ecosystem through a separate partnership with OpenAI. Under that agreement, Amazon will invest up to $50 billion, while OpenAI will use AWS infrastructure, offer its models through Amazon Bedrock and provide Amazon access to its AI models for use across the company's products and services.