AWS re:Invent 2024: Amazon takes pole position in AI race

OpenAI barely got a mention during the opening keynote address at AWS re:Invent in Las Vegas on Tuesday but was clearly in the crosshairs of a new arsenal of AI weapons announced. Matt Garman, the new CEO of AWS, welcomed his predecessor, Amazon CEO Andy Jassy, on stage to make the big announcement himself.

Jassy introduced Amazon Nova, described as “a new generation of foundation models (FMs) that have state-of-the-art intelligence across a wide range of tasks, and industry-leading price performance”.

The Nova models will be available on the Amazon Bedrock platform to enterprise customers and developers, and are not envisaged as direct competition to the likes of ChatGPT and Google Gemini – or at least the free versions of those products. Coming at a time when OpenAI is intensively courting enterprise customers, the new models are geared to regain the initiative in the fastest growing category of cloud computing – an industry AWS has dominated until now.

The models include:

• Amazon Nova Micro – a very fast, cost-effective, text-to-text model;
• Amazon Nova Lite – a low-cost multimodal model that processes image, video, and text inputs;
• Amazon Nova Pro – a multimodal model with “the best combination of accuracy, speed, and cost for a wide range of tasks”; 
• Amazon Nova Premier – “the most capable of Amazon’s multimodal models for complex reasoning tasks and for use as the best teacher for distilling custom models”.
• Amazon Nova Canvas – which generates studio-quality images; and
• Amazon Nova Reel – which generates studio-quality videos.

Amazon Nova Premier will be available in the first quarter of 2025, while the rest are now available.

Jassy displayed detailed lists of benchmarks, much the way smartphone manufacturers display benchmarks and specs that show their handsets are superior to the rest.

“If you compare Nova Lite to OpenAI’s GPT 4.o, you can see that we’re equal or better on 17 of the19 benchmarks; equal or better on 17 and 21 benchmarks versus Gemini,” he said.

“There’s other things that I think you’re going to really like about these models. First, they’re really cost effective. They’re about 75% less expensive than the other leading models. Two, they are fastest models that you’ll see with respect to latency.”

He enthused about Nova Canvas, which allows users to input natural language, and get studio quality images back, which can then be edited.

“It allows you to edit images using natural language text inputs. It gives you controls for colour scheme and layout. It has a number of built in controls for responsible use of AI, including watermarking for traceability as well as content moderation.”

He was equally enthusiastic about Nova Reel: “It gives you full control of camera, you have motion control, it lets you do 360-degree rotation, it lets you of zoom. It also has built-in AI controls for safe AI and content moderation.”

He said it would be launched with the ability to do six second videos, “which works really well for a lot of marketing and advertising”, up to two-minute videos in the next few months.

“What’s going to be next for us in November? First thing is the team is going to be working really hard over the next year on the second generation of these models. In the Q1 timeframe, we anticipate giving you a speech-to-speech model which will allow you to input speech and get speech back. Then around mid-year, we’re going to give you multimodal-to-multimodal. You’ll be able to input text speech, images, video and output text speech, images and video.”

He described these as “any-to-any” modality capabilities,

“This is the future of how frontier models would be built.”

All Amazon Nova models are integrated with Amazon Bedrock, a fully managed service that allows customers to experiment with and evaluate these as well as other foundation models, to determine the best model for an application.

The Nova models are also integrated with Amazon Bedrock Knowledge Bases, which allows the use of Retrieval Augmented Generation (RAG) to enable customers to ground responses in their own data.

Companies already using the tools include digital marketing company Dentsu Digital, which is integrating Amazon Nova Reel into its creative process, enabling its team to improve and accelerate the development of its campaigns – from briefing to concept development, to creative video content generation. Reel reduces the overall time it takes to generate new assets from weeks to days.

Musixmatch, the world’s largest lyrics platform with over 80-million users and a database of more than 11-million unique lyrics, has integrated Reel with Musixmatch Pro, which helps creators distribute lyrics across all the major digital streaming services and social networks. Emerging artists can use Reel to produce high-quality music videos using their songs’ context as inputs, and customise them with natural language prompts.

• To get started with Amazon Nova models, visit: https://aws.amazon.com/nova/

* Arthur Goldstuck is CEO of World Wide Worx and editor-in-chief of Gadget.co.za. Follow him on Bluesky on @art2gee.bsky.social.