Amazon on Tuesday unveiled a suite of artificial intelligence models in its boldest move yet to compete with tech giant rivals in the fast-growing generative AI sector.
The launch of its own line of foundation models marks Amazon’s latest push to strengthen its position against forerunners Microsoft, Google, Meta and OpenAI, the creator of ChatGPT.
Read more: King Charles III hosts emir of Qatar on UK state visit
Until now, Amazon’s AI offerings through its AWS cloud service had largely been limited to providing access to models from other companies, including Anthropic, an AI startup it backs.
Even if Google, Microsoft and OpenAI have taken the lead on AI, AWS remains the market leader in cloud computing, which is needed to power artificial intelligence tools and products.
“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,” said senior vice president Rohit Prasad, who is leading the company’s AI efforts.
“Our new Amazon Nova models are intended to help with these challenges,” he added.
Read more: South Korea’s President Yoon reverses martial law after lawmakers defy him
The Amazon Nova family includes six AI models handling tasks from text creation to video generation.
The company says the models are at least 75 percent cheaper than comparable offerings available on AWS servers and faster than similar models.
The initial lineup includes Nova Micro for fast text processing, Nova Lite for basic multimedia tasks, and Nova Premiere, set for an early 2025 release, for complex reasoning.
Supporting 200 languages, the models can be customized using customers’ proprietary data – a feature Amazon hopes will attract enterprises developing specialized AI applications.
Two dedicated models target creative content: Nova Canvas for image generation and Nova Reel for video creation.
Amazon emphasized built-in safety measures for the new offerings, which will be available through AWS’s Bedrock service, with usage guidelines detailing specific use cases and limitations.