Amazon is reportedly developing a new, more powerful generative AI model under the codename “Metis” to compete directly with top-tier systems from OpenAI and Google. According to internal documents and sources familiar with the matter, the project has the direct attention of Amazon CEO Andy Jassy and aims to create a successor to the company’s current “Titan” family of models.
The “Metis” model is being positioned as Amazon’s next flagship AI, designed to be a state-of-the-art system with advanced capabilities. It is expected to be multimodal, meaning it can process and generate responses based on a combination of text and images. More significantly, reports suggest Metis will incorporate agent-like functionality. This would allow the AI to perform complex tasks, make decisions, and automate sequences of actions on behalf of a user, such as booking travel or optimizing business processes.
This development signals a major strategic push by Amazon to reclaim ground in the high-stakes AI race. While the company’s cloud division, Amazon Web Services (AWS), offers access to various third-party models like Anthropic’s Claude through its Bedrock platform, developing a cutting-edge proprietary model is crucial for attracting and retaining enterprise customers.
The new model is reportedly powered by AWS’s own custom AI chips, Trainium and Inferentia, reducing reliance on external hardware providers like Nvidia. By integrating retrieval-augmented generation (RAG), Metis is also being trained to pull real-time information from the web to provide more accurate and up-to-date answers. While no official release date has been announced, the project underscores Amazon’s ambition to be a leader, not just a facilitator, in the generative AI landscape.
English 
Deutsch