Amazon It has iroduced a series of artificial ielligence models under the new brand name ‘Nova’. These models will be available as part of the Amazon Bedrock suite on AWS.
According to Amazon’s stateme, the three main models of this collection include the following:
- A text model named Amazon Nova Micro which is optimized in terms of speed and cost.
- An “extremely low-cost” multimodal model named Amazon Nova Lite which can receive and process images, video and text.
- A “fully capable” multimodal model named Amazon Nova Pro.
The tech gia is also training another model named Amazon Nova Premier It is referred to as Amazon’s “most powerful multimodal model” for complex reasoning tasks. Amazon plans to release this model in early 2025.
Amazon Nova image and video production artificial ielligence model
There are two other models in the Nova series: Amazon Nova Canvas To produce the image and Amazon Nova Reel For video production. Amazon says the two models are “watermark-capable,” promoting the “responsible use of artificial ielligence.”
The following video ad for a pasta brand was created by model Nova Reel:
Finally, Amazon is preparing a speech-to-speech model and a “multimodal to multimodal native model” for late 2025.
Amazon has iroduced these new models at the AWS re:Inve conference, which is currely being held in Las Vegas. At the conference, the tech gia also announced that it is building a massive AI computing cluster powered by its Trainium 2 chips. It has also partnered with Ahropic to build it, which recely invested $8 billion in two rounds.
Amazon’s stateme about this computing cluster says:
“Once complete, the cluster is expected to be the world’s largest AI computing cluster to date available to Eropic to build and deploy its future models on.”
