Amazon has been reportedly wanting to participate in AI’s “inference” model game.
According to Business Insider, Amazon is developing AI models that incorporate advanced “inference” capabilities, similar to models such as Openai’s O3-Mini and China’s AI Lab Deepseek’s R1. This model could soon be released in June under Amazon’s NOVA brand. The brand was introduced last year at RE: Invent Developer Conference.
The inference model takes a more step-by-step, more consideration approach to answering queries. This tends to increase reliability in domains such as mathematics and science.
According to the report, Amazon aims to embrace a new model of “hybrid” inference architecture, along with the recently released Claude 3.7 Sonnet line. If that is achieved, the model can provide quick answers and more complex, extended thinking within a single system.
Amazon also wants NOVA’s inference model to be price-efficient than its competitors, Business Insider argues. It could be a tall order. Deepseek has built a reputation for pricing models very cheaply.