Amazon announces its own set of Nova AI models

The company is also training a model called Amazon Nova Premier, which it says will be “our most efficient multimodal model for complex inference tasks.” Amazon aims to make the Nova Premier available “in early 2025.”

Amazon also provides content generation models: Amazon Nova Canvas – an image generation model and Amazon Nova Reel – a video generation model. The company says these models have “watermarking capabilities” to “promote the responsible use of artificial intelligence.” As an example of what is possible with Nova Reel, Amazon shared this false advertisement for a fake brand of pasta.

Later in 2025, Amazon plans to release a speech-to-speech model and a “native multimodal to multimodal model”, according to the blog post.

Amazon announced these novel models at the AWS re:Invent conference, currently taking place in Las Vegas. During the show, the company also said that it is building a massive AI computing cluster based on Trainium 2 chips in partnership with Anthropic (in which it has invested $8 billion). “When completed, it is expected to be the world’s largest artificial intelligence computing cluster to date.” reported so far, and which will be available to Anthropic on which it will be able to build and implement its future models” – according to Amazon.

The company, like many other substantial tech players, is racing to release novel AI products and features to overtake newer companies like OpenAI. Amazon’s advantage may be that much of its internet infrastructure is already powered by AWS – huge enterprises may be more willing to take advantage of Amazon’s AI offerings because the company already has a trusted reputation. Even the CEO of Apple appeared on stage today at re:Invent to talk about how the company relies on Amazon’s custom AI chips.