Amazon Web Services (AWS) has significantly expanded its artificial intelligence offerings, unveiling the new Nova 2 family of AI models and a groundbreaking service called Nova Forge. Announced during CEO Matt Garman's keynote at the annual AWS Re:Invent conference, these innovations empower enterprise customers with advanced generative AI capabilities and unprecedented customization options.

Building on the success of the original AWS Nova models, first introduced at last year's Re:Invent, the updated Nova 2 family represents a substantial upgrade. The initial release included four text-generating and one image-generating model. Garman highlighted the rapid adoption, stating, "The momentum has been really fantastic. Nova has grown to be used by tens of thousands of customers today, everyone from marketing giants to tech leaders like Infosys or Blue Origin or Robinhood to innovative startups like NinjaTech AI, and today, we're making Nova even better."

Introducing the Nova 2 AI Model Family

The Nova 2 family introduces four distinct models, each tailored for specific AI applications:

  • Nova 2 Lite: A cost-effective reasoning model designed for everyday tasks, capable of processing text, images, and videos to generate text. Reasoning AI models "think" before responding, enhancing output quality.
  • Nova 2 Pro: An advanced reasoning agent built for highly complex tasks, such as coding. It processes text, images, videos, and speech.
  • Nova 2 Sonic: A novel speech-to-speech model optimized for conversational AI applications.
  • Nova 2 Omni: A powerful multimodal reasoning and generation model that accepts input across images, text, video, and speech, producing both text and images.

Nova Forge: Custom AI for Enterprises

Complementing these model enhancements, AWS introduced Nova Forge, a new service enabling cloud customers to develop their own "frontier" versions of AWS Nova models, dubbed "Novellas." This bespoke service, reportedly costing $100,000 annually according to CNBC, provides enterprises with access to pre-trained, mid-trained, or post-trained models, which they can then further refine using their proprietary data.

Garman emphasized that Nova Forge addresses a critical challenge faced by enterprises: the degradation of core reasoning when existing AI models are extensively customized with new, proprietary data. He explained:

"The more you customize models, the more you add a bunch of data in post-training, these models tend to forget some of that interesting stuff that it learned earlier—the core reasoning."

He drew an analogy to human language acquisition:

"It's a little bit like humans trying to learn a new language. When you start when you're really young, it's actually relatively easy to pick up, but when you try to learn a new language later in life, it's actually much, much harder. Model training is kind of like this too."

Early adopters of the Nova Forge service include prominent companies such as Reddit, Sony, and Booking.com.