Full Title: Anthropic Is Neck-and-Neck With Rivals After Its Latest Release
Highlights
Who’s it for? Claude has garnered a reputation as the most literary and intellectual model on the market — the go-to choice for those working on creative or writing-intensive projects. It also caters to firms by letting them fine-tune its models for specific purposes. (View Highlight)
What’s with the name? The company usually divides its models into threes: There’s the compact, efficiency-oriented Haiku, the balanced Sonnet, and the powerhouse Opus. In this case, it’s releasing only a new Sonnet model, with Haiku and Opus presumably coming soon. (View Highlight)
What’s different? Most of the changes are relatively subtle: Claude is now better at math, reasoning, and code generation, among other skills. It’s also said to be 80% cheaper and twice as fast as Claude 3 Opus, its previous state-of-the-art model. (View Highlight)
One notable addition is the new Artifacts feature, which lets you tweak text and images directly in the Claude platform itself. Most chatbots are relatively minimalistic, so this could be one of the first bridges between traditional chatbots, word processors, image editors, and other types of interfaces. (View Highlight)
The cost of training AI platforms is doubling every nine months, with little sign of leveling off, according to a recent analysis from the research organization Epoch AI. At this rate, training new models could soon cost more than $1 billion, especially when electricity and hardware expenses are factored in. And that’s without considering employee compensation, which can account for up to half of the total cost of training a new model. (View Highlight)
Whether this trend will continue into the next decade remains an open question. Some analysts think AI companies will soon run out of data — and that both training costs and performance gains will start to plateau as a result. Others believe that as demand for AI chips goes through the roof, they’ll get even more expensive. That would likely make it more difficult for smaller startups to keep up with their higher-profile rivals. (View Highlight)
(View Highlight)