Key takeaways
Amazon Web Services has launched a comprehensive expansion of its AI capabilities, introducing four new Nova 2 models alongside pioneering services that allow enterprises to build custom AI models and deploy automated browser agents.
The announcements, made at the company's re:Invent conference in Las Vegas on December 2, position AWS to compete more aggressively with OpenAI, Google, and Anthropic in the enterprise AI market.
The Nova 2 family includes four distinct models designed for different levels of complexity and use cases.
Nova 2 Lite serves as a cost-effective reasoning model for everyday tasks like customer service chatbots and document processing.
Nova 2 Pro represents Amazon's most advanced reasoning model, capable of handling complex tasks, including agentic coding and long-range planning.
Nova 2 Sonic provides speech-to-speech capabilities with support for multiple languages, while Nova 2 Omni marks an industry first as a unified multimodal model that can process text, images, video, and speech inputs while generating both text and images.
According to benchmarking data released by Amazon, Nova 2 Pro performs equal to or better than competing models on multiple tests.
The model matches or exceeds Anthropic's Claude Sonnet 4.5 on 10 out of 16 benchmarks, OpenAI's GPT-5.1 on 8 out of 16 benchmarks, and Google's Gemini 3.0 Pro Preview on 15 out of 19 benchmarks.
Building custom AI models with Nova Forge
Nova Forge represents a significant departure from traditional AI model development by giving enterprises access to pre-trained, mid-trained, and post-trained model checkpoints.
This "open training" approach allows companies to integrate their proprietary data at various stages of the training process, creating what Amazon calls "Novellas" – custom frontier models that blend organizational knowledge with Amazon's base capabilities.
"We are a frontier lab that has focused on customers," Rohit Prasad, Amazon's head scientist for artificial general intelligence, told CNBC. "Our customers wanted it. We have invented on their behalf to make this happen."
The service costs $100,000 annually and aims to provide a more affordable alternative to building AI models from scratch, which can cost hundreds of millions or even billions of dollars.
Prasad indicated that tens of thousands of organizations are using Nova models each week, though he noted that Anthropic's models remain the most popular on AWS Bedrock, with Nova ranking as the second-most used model family.
Early customer deployments show tangible results
Reddit has emerged as a prominent early adopter of Nova Forge, using the service to overhaul its content moderation systems.
The platform needed an AI model sophisticated enough to analyze potentially offensive content that standard commercial models would refuse to process.
"Working with Nova Forge is allowing us to improve content moderation on Reddit with a more unified system that's already delivering impressive results," said Chris Slowe, CTO of Reddit, according to Amazon's announcement.
"We're replacing a number of different models with a single, more accurate solution that makes moderation more efficient. The ability to replace multiple specialized ML workflows with one cohesive approach marks a shift in how we implement and scale AI across Reddit."
Other companies deploying Nova Forge include Booking.com, Cosine AI, Nimbus Therapeutics, Nomura Research Institute, OpenBabylon, and Sony.
Each organization is building custom models tailored to its specific industry requirements and data environments.
Nova Act automates browser-based workflows
Alongside the new models, AWS introduced Nova Act, a service designed to build and manage AI agents that automate browser-based tasks.
Powered by a custom Nova 2 Lite model trained through reinforcement learning on simulated web environments, Nova Act achieves 90% reliability on early customer workflows.
The service enables automation of tasks such as updating customer relationship management systems, testing website functionality, and submitting insurance claims.
Hertz used Nova Act to accelerate its software delivery by 5x, transforming quality assurance processes that previously took weeks into hours. The rental company processes millions in daily bookings through its platform, making the efficiency gains particularly significant.
Password management company 1Password integrated Nova Act to reduce manual login steps across hundreds of websites, while startup Sola Systems deployed the service to automate hundreds of thousands of workflows per month for clients across business-critical tasks.
Strategic positioning in the competitive AI market
The announcements reflect Amazon's broader strategy to differentiate itself in the intensifying enterprise AI market.
While Microsoft has leveraged its OpenAI partnership and Google pushes its Gemini models across cloud services, Amazon is betting on customization and control as key differentiators.
"A year ago, there were questions about whether we'd missed the wave, but now, most people are building their production systems in AWS because of what we've built over the past couple of years," AWS CEO Matt Garman told Axios. "People are now realizing that Amazon has a great platform for AI."
The Nova 2 models are available immediately through Amazon Bedrock, the company's AI service platform.
Nova Forge is currently available in the US East (N. Virginia) AWS Region, with additional regional availability planned for the coming months.
Early access to Nova 2 Pro and Nova 2 Omni is being provided to Nova Forge customers, allowing them to test Amazon's most advanced models alongside internal Amazon teams.
Amazon also announced the availability of Trainium3 UltraServers, which deliver approximately four times more compute performance and energy efficiency compared to previous generations, supporting the infrastructure requirements for training and deploying these advanced AI models.
Read more:
Thai Singer Files Complaint Over Deepfake Pornography
Oxford University trials new AI gateway service from Microsoft and Cloudforce
Google Launches AI Textbook Tool Showing Improved Student Retention