In this photo illustration, the Claude AI logo is seen on the smartphone and the human slogan on the computer screen. (Clarification of photos by Pavlo Gonchaar/SOPA photos/Lightrockwal via Getty Images)
SOPA photos | Lightrockket | Gety pictures
Man, and AmazonOn Thursday, the Openai opponent launched its strongest group artificial intelligence Models so far: Claude 4.
The company said that the two models, called Claude Obus 4 and Claude Sonnet 4, define a “new standard” when it comes to artificial intelligence employees and “can analyze thousands of data sources, carry out long -term tasks, write high -quality content, and perform complex procedures, according to the version.
Anthropor, which was established by former Openai Research, launched Claud Chatbot in March 2023. Since then, it has been part of the growing male arms race between startups and technology giants alike, a market from which to do so. The highest 1 trillion dollars In revenues during a decade.
Companies in each industry seems to be accelerated to add Chatbots and agents operating in Amnesty International to avoid them To leave behind it By competitors.
Antarubor stopped investing in chatting at the end of last year, and instead focused on improving the ability of Claude to carry out complex tasks such as research and coding, and even writing the rules of complete symbol, according to Garrid Kaplan, chief science employee in man.
He also admitted that “the more complicated, the more risk is that the model will start from bars … We really focus on treating this so that people can delegate a lot of work simultaneously to our models.”
“We have trained these models since last year and really expected,” Kaplan said in an interview. “I think these models are much stronger as agents and milestagine. It was definitely an internal struggle just because some of the new infrastructure we used to train these models … made it very useful for the teams in terms of getting everything and operating it.”
Kaplan added that as soon as they tried to get the model test to stop the new models because they need to be repeated, no one wants to give up access.
Anthropor said that Claude Obus 4 was “the best coding model in the world” and could work independently for a full day working day – seven hours.
Both models can search in the web to complete the tasks on the user and alternative between thinking and using the tool, according to Anthropic. The company also said that when it allows access to local files, they can extract and save “the main facts to maintain continuity and build implicit knowledge over time.”
“I do a lot of writing with Claude, and I think before OPUS 4 and Sonnet 4, I used to use models mostly as a thinking partner, but I still do most of myself,” said Mike Crager, chief product employee in Anthropor in an interview. “They crossed this threshold where most of my writing is in fact … OPUS mostly, and it is now unknown from my writing.”
“I like to pay a kind of border on two sides. Like one of them is a piece of coding and conscious behavior in general, and this works to operate many of the startups in these coding … but also, we push the borders on how to learn these models actually from the really useful writing partner, too.”
The annual revenues of the anthropologist reached $ 2 billion in the first quarter, as confirmed by the company Last weekMore than twice the price of one billion dollars in the previous period. Kate Ginsen, head of revenue, said in an interview with him recently with CNBC that the number of customers spends more than 100,000 dollars annually with anthropor. Jump A year ago.
Wall Street continues to pour money into emerging companies from artificial intelligence, such as Antarbur: The company got 2.5 billion dollars and five years Credit line Last week, to increase its liquidity in the medium artificial intelligence competition.
https://image.cnbcfm.com/api/v1/image/107431256-1718898592166-gettyimages-1737213579-PGONCHAR_4313.jpeg?v=1747931481&w=1920&h=1080
Source link