At its re:Invent 2024 conference, Amazon’s AWS unveiled Nova, a cutting-edge family of AI models designed to revolutionize how businesses leverage AI. This announcement marks a significant leap in Amazon’s AI development, spotlighting six generative AI models designed to simplify text, image, and video content generation. These AI tools are accessible to enterprise clients via Amazon Bedrock.
“Inside Amazon, we have about 1,000 Gen AI applications in motion, and we’ve had a bird’s-eye view of what application builders are still grappling with,” Amazon SVP of Artificial General Intelligence Rohit Prasad said in a statement. “Our new Amazon Nova models are intended to help with these challenges for internal and external builders, and provide compelling intelligence and content generation while also delivering meaningful progress on latency, cost-effectiveness, customization, information grounding, and agentic capabilities.”
Meet the Nova Models
The Nova series includes five currently available models and a sixth, Nova Premier, slated for release in Q1 2025. The AI models are built to address diverse enterprise needs, from text to advanced multimedia content generation.
Text-only models include Micro, Lite, and Pro:
- Nova Micro: Ideal for quick text-to-text tasks, Micro is the smallest in the lineup but boasts the lowest latency. Its context window of 128K tokens ensures efficiency for tasks such as text summarization, translation, content classification, interactive chat and brainstorming, and simple mathematical reasoning and coding.
- Nova Lite: The model accepts images, videos, and text as inputs but only generates text. It processes inputs up to 300K tokens in length and can analyze multiple images or up to 30 minutes of video in a single request.
- Nova Pro: The most robust among the trio, Pro delivers superior accuracy and multitasking capabilities. It also features a 300K-token capacity, making it suitable for complex workflows.
Creative content models include Canvas and Reel:
- Nova Canvas: Specializing in image creation, Canvas can generate visuals based on text or image prompts. It’s designed for industries like advertising and entertainment, with tools for color adjustments and background edits.
- Nova Reel: A video generation powerhouse, Reel produces short clips from text and image prompts. Users can manipulate camera movements using natural language commands, making it an ideal tool for marketers and content creators.
- Nova Premier: The most advanced model, built for complex reasoning and fine-tuning custom AI solutions. Amazon Nova Premier is still in training. It’s set to launch in early 2025.
What Makes Nova Different?
One standout feature is Nova’s focus on cost efficiency. Amazon Nova Micro, Amazon Nova Lite, and Amazon Nova Pro are at least 75 percent less expensive than the best-performing models in their respective intelligence classes in Amazon Bedrock, Amazon said. They also excel at customization, allowing businesses to fine-tune models using their proprietary data for improved accuracy and relevance.
In 2025, some Nova models’ context windows will expand to support over two million tokens, the equivalent of more than 1.5 million words, marking another leap forward. Additionally, AWS plans to launch a speech-to-speech model in early 2025, promising natural-sounding transformations by interpreting tone and cadence. An ambitious any-to-any model, combining inputs and outputs across text, speech, images, and video, is expected later that year.
Learn more about the other top companies redefining AI and advancing this dynamic technology.