Amazon takes a new step in the race for artificial intelligence (AI) against the big names in the sector. Until now, the American giant only offered generative AI software from other companies on its Amazon Bedrock platform, notably those from Meta, Mistral AI or Anthropic. Amazon had also announced at the end of November that it would invest an additional 4 billion dollars in the generative artificial intelligence start-up Anthropic, thus bringing its total injections to 8 billion.
However, on Tuesday, its remote (cloud) computing subsidiary, Web Services (AWS), presented its own generative AI models. The interface has been called Amazon Nova and is available in six versions, which range from generating text from a written query (Micro) to creating images (Canvas) and videos (Reel). It was presented at the AWS re:Invent conference, organized in Las Vegas (Nevada).
Artificial intelligence takes action
And Amazon assures that the three least advanced versions of Nova are “at least 75% cheaper than the best performing models in their category on Amazon Bedrock” and are also “the fastest”.
Not directly accessible to individuals
Unlike OpenAI’s ChatGPT or Microsoft’s Copilot assistant, Nova is only accessible to businesses and AWS client developers and not directly to individuals. Amazon plans to launch two new versions of Nova in 2025.
One called multimodal to multimodal will be capable of processing text, image, video or sound and producing a result in the chosen form, written, illustrated or sound. The second will be of type « speech to speech »that is to say that a voice request will produce an audio response, on the principle already seen for virtual assistants like Apple’s Siri or Amazon’s Alexa. Nova models can operate in 200 different languages and be customized to customer needs through the Bedrock platform.
“Within Amazon, we have around a thousand generative AI applications in operation, which has given us an overview of the difficulties that app developers still have” to create their own software, Amazon’s Rohit Prasad was quoted as saying in the presentation.
“Our Amazon Nova models are made to help overcome these challenges”continued the vice-president in charge of artificial general intelligence (AGI). AGI aims to develop an AI with intelligence equivalent to that of a human being.
(With AFP)
