What is new in the Human 4 Gen Ai models?

The latest versions of the anthropor Claude Temporary artificial intelligence models It appeared for the first time on Thursday, including a heavier model in service designed specifically for coding and complex tasks.

I launched the new anthropologist Claude 4 Obus and Claude 4 Sonit The models said during its blog with the Claude conference for developer and executives that the new tools define a big step forward in terms of thinking and deep thinking skills.

The company launched the previous model, Claude 3.7 Sonnet, in February. Since then, artificial intelligence developers have increased their game. Openai GPT-4.1 was released in AprilFocusing on an expanded context window, along with the new O3 Thinking Form family. Google followed in early May with An updated version of Gemini 2.5 Pro He said he was better in coding.

CLAUDE 4 OPUS is a larger and more intense resource model designed to deal with special challenges in particular. Anthropor CEO, Dario Amani, said that test users have seen that he was dealing quickly with the tasks that may have taken several hours to complete.

He said during “in many ways, as we often find with large models Major.

Claude 4 Sonnet is a smaller model, with improvements based on the Claude 3.7 Sonnet model. Amani said the 3.7 model often had problems with excessive opinion and sometimes doing more than the person asked to do. He said that while it is a less dense resource model, it is still a good performance.

“In fact, it is also also on some coding criteria, but I think it is focused in a more narrow way,” said Amoudi.

Anthropor said the models have a new ability, still conducting a experimental test, as they can use tools such as web searches while participating in expanded thinking. Models can alternate between thinking and use tools to get better responses to complex queries.

Models provide semi -fixed response conditions and expanded thinking patterns.

All paid plans offer OPUS and Sonnet models, while the free plan has only Sonnet model.

The new models show the human focus on building strong coding models. “Claude models have created a strong leadership in the field of software engineering and the latest Claude 4 version that extends that leadership.”

Anthrop works on safety protocols with new Claude models

Upon launching the Claude Obus 4 model, Anthropor said it was taking increasing safety precautions to reduce the risk of misuse of Claude. in Blog postThe company said it did not specify whether the model already required to protect the ASSL-3 standard, but it does this as a preventive measure.

Safety reserves are specially designed to prevent Claude from helping to develop chemical, biological, radiological or nuclear weapons. Anthropor said it was limited attacks known as the global courage that allows the attackers to circumvent the current protocols. “We will continue to evaluate the CBRN capabilities of the CLADE OPUS 4,” said Anthropor’s blog. “If we conclude that Claude Obus 4 did not exceed the relevant power threshold, we can remove or adjust the ASL-3 protection.”

Chandricicaraan said that the implementation of safety standards is worth referring to. He said: “This includes reinforced cybersecurity measures and rapid works to mitigate the risks associated with strong artificial intelligence systems.” He said that the new models show the company’s focus on the budget of the new technology with safety.