Claude Sonnet 4.5 is Anthropic’s safest AI model yet

Cyber Security, ICT, Most Popular, Trends News

No Comments

Photo of author

By Karla T Vasquez

WhatsApp Group Join Now
Telegram Group Join Now


In May, the ethnographic announced two new AI systems, Opus 4 and sonnet 4The Now, less than six months later, the company is launching Sonnet 4.5 and called it the best coding model in the world to today. The basis of anthropic for this claim is a selection of benchmarks where the new AI is not just its predecessor, but also the more expensive OPS 4.1 and Google also surpasses competitive systems Gemini 2.5 Pro And GPT -5 to OpenThe For example, in Osworld, a suit that tests AI models in real-world computer tasks, sets a record score of 4.5 61.5 percent, which holds 5 percent percent points above OPS 4.1.

At the same time, the new model is capable of working autonomously on multi-step projects for more than 30 hours, can maintain a significant improvement of OPAS 4 at seven or seven hours. This is an important milestone for an agent system that wants to create anthropic.

Sonnet 4.5 exceeds the old models of anthropic in coding and agent functions.

Sonnet 4.5 exceeds the old models of anthropic in coding and agent functions.

(Ethnographic)

Probably more importantly, the company claims that Sonnet 4.5 is the safest AI system till today, with the model “broad” security training with the model. This training translates a chatboat anthropological “Psychophyse, cheating, energy looking for finding and confusing thoughts”-“less prone” —the potential model features that contain Open AAE in hot water in recent monthsThe At the same time, the Sonnet 4.5 has strengthened the protection of Sonnet 4.5 against anthropological instant injection attacks. Due to the sophistication of the new model, the ethnic is publishing Sonnet 4.5 under its AI protection level 3 Framework, which means that it comes with filters designed to prevent possible dangerous results related to chemicals, biological and nuclear weapons.

Sonnet 4.5 is a chart that shows how the security tests compare it with other border models.

Sonnet 4.5 is a chart that shows how the security tests compare it with other border models.

(Ethnographic)

With today’s announcement, anthropological Claud Product is also turning the quality of life throughout the stack. To begin with, the company’s popular coding agent Clod Code has a refresh terminal interface, with which checkpoints include a new feature. You can probably guess from the name, they let you save your progress and go back to the previous kingdom if Claud writes some fancie code that doesn’t work as much as you imagine. File creating, which ethnographic Roll out at the beginning of the monthNow directly available in the conversation with the chatboat and if you join the waitlist clode for Chrome, you can start using the extension today.

The API price for Sonnet 4.5 has the same amount of output token of one million input token per $ 3 and $ 15. Sonnet 4.5 closes a strong September for the release anthropological. Just one day after Microsoft Copilot has been added to CLOD MODE AT 365 Last week, Open Admission Its rival provides the best AI for work related to work.

Leave a Comment