Amazon’s new, AI-powered Alexa+ is reportedly leaning heavily on Anthropic’s large language model, Claude, for many of its complex tasks, according to sources familiar with the matter, despite Amazon’s public denial of these claims.
Earlier this week, Amazon unveiled its long-anticipated update to its decade-old Alexa devices. This new version, “Alexa+,” will introduce a subscription model, costing $19.99 per month (or free for Amazon Prime members). The service is slated to launch in early access next month.
During a demonstration, Alexa+ showcased its ability to perform tasks such as making dinner reservations, ordering groceries, and booking rides with services like Uber—capabilities largely unavailable in prior versions. This upgrade comes as Alexa faces increasing competition from generative AI chatbots like OpenAI’s ChatGPT, who have rapidly expanded beyond text-based interactions into AI-generated audio, images, and video.
Reports indicate that Anthropic’s Claude model is handling the majority of customer queries for the new Alexa. One source, speaking anonymously because the information is confidential, suggested that Claude is being utilized to manage the more complex requests, those demanding greater “intellectual heft.”
In response to these claims, an Amazon spokesperson stated that information provided in the story is “false.” Furthermore, Amazon representatives have said that, in the last four weeks, their “Nova” platform has autonomously handled over 70% of Alexa conversations, including complex user requests.
“That said, from a customer perspective this doesn’t matter — both are excellent models and in there to deliver the best experience for customers,” the spokesperson added. Amazon has designed a “best model for each task” approach, ensuring Alexa+ uses whichever AI platform is most appropriate for the task at hand.
Amazon has agreed to invest a substantial $8 billion in Anthropic, a company founded by former OpenAI employees. Amazon’s CEO, Andy Jassy, described the Alexa overhaul as a “rearchitect[ing]” of the assistant’s core.
In addition to their financial commitment, Amazon has developed its own AI models, introducing the Nova series late last year. Amazon offers Anthropic’s Claude model to its customers as part of its Amazon Web Services Bedrock offering, which provides access to multiple AI solutions. This allows users to choose between Amazon’s Nova and Titan models, alongside options from Mistral and other providers.
While Amazon claims to use Bedrock to power Alexa, sources indicate that Claude has been handling the more complex tasks demonstrated at the recent device event in New York. These tasks require deeper reasoning and thought.
As part of its initial investment in Anthropic, Amazon gained access to a specific amount of Anthropic capacity without charge for a duration of 18 months, according to one source. However, this agreement has expired, and the companies are currently renegotiating the terms. Another source added that Anthropic’s model has made inroads far beyond Alexa within Amazon, assisting in product search and advertising as well.
Panos Panay, Amazon’s senior vice president of devices and services and the leader of Alexa’s redesign, described Anthropic as an “awesome” partner, noting Claude’s foundational model as being “incredible.” During an interview on Wednesday, Panay told CNBC “We use Amazon Bedrock—Alexa picks the right model to get the job done.”