Amazon’s New AI Reasoning Model
Amazon is reportedly preparing to launch its own AI model with advanced reasoning capabilities, according to a recent report. The model is slated for release in June, as part of the Nova series of generative AI models introduced in December 2024.
An AI reasoning model is designed to tackle complex problems by exploring multiple potential solutions and simulating the steps involved in problem-solving, often using techniques known as ‘chain-of-thought.’ Amazon’s approach, however, is described as a ‘hybrid reasoning’ system, capable of providing rapid answers while also being able to generate more in-depth and complex outputs once it has had time for more ‘thinking.’
This development could place Amazon in closer competition with Anthropic, an AI startup in which Amazon has already invested significantly. Amazon has invested approximately $8 billion in Anthropic.
Competitive Strategy
Amazon aims to position its Nova reasoning model competitively, particularly concerning pricing. The company is also targeting high rankings on external benchmarks used to assess AI models. These benchmarks, such as SWE, Berkeley Function Calling Leaderboard, and AIME, are designed to gauge an AI model’s ability to perform tasks related to coding and mathematics.
Rohit Prasad, an AI researcher, leads Amazon’s AGI division and is responsible for the development of the new reasoning model, as the report states. The news suggests an aggressive push by Amazon to strengthen its presence in the growing artificial intelligence landscape.
