AI startup Anthropic has announced Claude 3, a new family of AI models. This includes Opus, the company’s “most intelligent model,” which it says outperformed OpenAI’s GPT-4 on a variety of tasks.
This family includes two other models. One is the Sonnet, which is currently available on Opus, and the other is the Haiku, which is the least powerful of the three models, but Anthropic says this will be available later.
In a series of industry benchmark assessments including undergraduate and graduate level reasoning, mathematics, and general knowledge, Opus outperforms GPT-4 — the most powerful version of OpenAI’s ChatGPT — and Google’s Gemini.
Anthropic said Opus “is leading the frontier in general intelligence by demonstrating near-human levels of comprehension and fluency in complex tasks.”
Meanwhile, all three models have improved in analysis and prediction, content creation, code generation, and conversation in languages other than English, such as Japanese and French, Anthropic said.
Anthropic said the new model can be used for customer live chat, autocomplete and data extraction. According to the company, Haiku “allows him to read research papers packed with information and data in less than three seconds,” including charts and graphs.
The new model is also the first in Anthropic’s range of products that can handle visual formats such as photos, charts, and graphs, and enterprise users can now use the model to read PDFs and presentation slides. said Anthropic.
Anthropic said that unlike existing models, the Claude 3 family is “significantly less likely to refuse to respond to prompts” that could pose harm by getting too close to the system’s guardrails. The company is also adding citations to new models to demonstrate accuracy.
“Of course, no model is perfect, and I think this is a really important thing to say up front,” said Daniela Amodei, co-founder of Anthropic and former vice president of safety policy at OpenAI. He said: told CNBC. “We’ve worked very hard to make these models as functional and safe as possible. Of course, there will be a place for them to still make things up from time to time.”
Anthropic was founded in 2021 by Amodei and his brother Dario, who also worked at OpenAI. Aiming to develop a “safer” chatbot than Competition in the AI field.The startup counts Google and Amazon as backers and has closed its doors Total of approximately $7.3 billion across 5 funding rounds Over the past year, in September, Amazon announced: Invest up to $4 billion and acquire a minority stake in the company. Google followed up $2 billion investment During October.
