Anthropic Claude 3 takes the LLM crown
Anthropic has been on a tear lately, releasing its powerful, new Claude 3 family of large language models (LLMs) that have recently taken the crown from rival OpenAI (where the Anthropic brother-and-sister founding duo Dario and Daniela Amodei and their five other co-founders previously worked — that’s gotta sting!) as most powerful in the world.
There are three different versions each of Claude 3, costing more and offering more capabilities than the other, though the cheapest and smallest/least powerful, Haiku, has earned praise from users for underpricing and outperforming the competition — namely, OpenAI’s GPT-3.5 Turbo model, and even OpenAI’s larger and more powerful GPT-4, on some metrics.
Following the release of Claude 3 in early March 2024, Amazon quickly added some of the versions of Claude 3 to its Bedrock service on AWS, which allows AWS cloud customers to select between various LLMs that are proprietary like Anthropic’s and Cohere’s, as well as open source ones from Mistral and Meta’s Llama 2, to build their own third-party apps atop those models for customers or employees.
In that light, Amazon throwing more investment money at Anthropic makes a lot of sense: Anthropic is offering powerful tools that eclipse competitors, and Amazon gets to host them exclusively for its cloud customers (though it should be noted, anyone can access Claude 3 via its APIs and direct consumer-facing website).
Amazon clearly sees there being a major upside and possible revenue generation opportunity from convincing customers to A) use its AWS cloud services and/or S3 B) build AI apps with its Bedrock platform and C) do so using competitive and bleeding-edge models like Claude 3.
It is the latest front in the ongoing cloud services competition between AWS and Microsoft Azure. And with Microsoft a major investor of OpenAI, it makes total sense Amazon wants their own powerful AI startup in their proverbial pocket.
