Amazon Might Be Developing an In-House Reasoning AI Model

Amazon is reportedly working on a reasoning-focused artificial intelligence (AI) model. As per the report, this model will be released as part of the company’s Nova family of AI models. This will likely be an enterprise offering via Amazon Bedrock and Azure AI Foundry, instead of being a consumer-focused product. The large language model (LLM), once released, will compete with other reasoning models available in the market such as OpenAI’s o3-mini, Google’s Gemini 2.0 Flash Thinking, and the DeepSeek-R1.

Amazon Said to Be Building a Reasoning Model

According to a Business Insider report, the tech giant is building this reasoning model in-house from scratch. Citing unnamed people involved with the project, the publication claimed that Amazon is trying to add “hybrid reasoning” capability to this model.

Last month, Anthropic released the Claude 3.7 Sonnet AI model, which also offered hybrid reasoning. The term describes an LLM that can both provide standard intelligence responses almost instantaneously as well as provide the slower responses that utilise test time compute method to break down complex problems.

Amazon is reportedly planning to introduce this unnamed model in June. The tech giant’s main priority is said to be making the model cost-efficient. This is likely so that the model can remain competitive in a market that is quickly getting saturated with reasoning models.

While making the model cheaper is a priority, the tech giant reportedly does not want that to compromise its performance. Citing the source, the report claimed that Amazon also wants the model to rank in the top 5 models on third-party AI leaderboards.

The report likely means Amazon wants the reasoning-focused model to rank in the Chatbot Arena’s leaderboard. Formerly known as Large Model Systems Organisation (LYSYS), it is a crowdsourced platform where developers and AI users rank different models based on their experience with them.

Currently, Grok-3-preview ranks on top of the leaderboard, followed by GPT-4.5-preview and Gemini 2.0 Flash Thinking.

Notably, the report about Amazon’s reasoning-focused model comes just days after it unveiled Alexa+, the AI version of its voice assistant with new features such as contextual understanding, better control over smart home appliances, and more.