Anthropic Releases ‘Most Intelligent’ AI Model in Rivalry With OpenAI

(Bloomberg) -- Artificial intelligence startup Anthropic is releasing a new AI model that it calls its fastest and most capable yet in a rivalry with OpenAI.
Anthropic introduced Claude 3.5 Sonnet, an updated version of its Claude 3 model that it just released in March. The company said Thursday that the new model — the technology that underpins its popular chatbot Claude — is twice as fast as its most powerful previous version. Anthropic said in its evaluations, the model outperforms leading competitors like OpenAI on several key intelligence capabilities, such as coding and text-based reasoning. 

“This is exciting because this is now, based on our evaluations and benchmarks, the best and most intelligent model in the industry,” Daniela Amodei, Anthropic’s co-founder and president, said in an interview. “We are aiming for this to be the model of choice for enterprise businesses across the stack.”

Anthropic was started in 2021 by a group of former OpenAI employees. The company has emphasized its commitment to building AI safely and responsibly, which it says helps gain trust with business customers. Anthropic is competing with other AI developers like OpenAI, Alphabet Inc.’s Google and Microsoft Corp., which are also building out their enterprise businesses. Last month, OpenAI released a new model, GPT 4o, that it similarly said was more capable, faster, and cheaper, although it hasn’t yet widely released a key voice assistant feature.
Amodei said that because Anthropic is focused on developing products primarily for business customers, it wasn’t distracted by areas such as creating images that can be enticing for research, but not practical for companies. 

Related:How to Monitor AI with AI

For example, Amodei said, “We can do image outputs from a research perspective, but drawing visuals of a snowboarding cat is not what our enterprise customers are asking us for, so we're not prioritizing that.”

Claude 3.5 Sonnet will be available for free on mobile and desktop up to a certain limit, while paying Claude subscribers can access the tool more frequently. The company has been testing the updated technology with a small group of users.
As an example of how customers are using Anthropic, Amodei pointed to pharmaceutical giant Pfizer Inc., which she said is using the chatbot to help with the research and development of drug discovery.

“The more intelligent the model is, the more capable it is of sort of supporting that kind of very high level intellectual work,” Amodei said.

The company is also releasing a new feature called Artifacts that will let users collaborate in real-time on coding, documents and other content generated by the Claude chatbot.