Anthropic has introduced a new AI model. It can doubt the accuracy of its answers

AI startup Anthropic has released an updated version of its most powerful artificial intelligence model, Opus 4.8. According to the company, early testers of the model noted that it is more likely to point out cases where it is unsure of the accuracy of its answers. Opus 4.8 also performs better at programming tasks on behalf of users, financial analysis and tasks that mimic the human reasoning process, Anthropic said.

The new model came out just a month or so after the previous release, Bloomberg notes. This is how the company aims to maintain an advantage over competitors, including OpenAI and Google, the agency believes. In recent months, Anthropic has demonstrated strong dynamics in the segment of AI programming tools, as well as gaining popularity among consumers amid the conflict with the Pentagon over AI security guarantees, Bloomberg recalls.

According to him, Anthropic may in the coming weeks open wider access to new AI models with cybersecurity testing capabilities comparable to the Mythos model, which previously raised concerns among experts because it could be used by hackers. Because of this, the startup gave access to Mythos only to a limited number of researchers. Anthropic itself previously reported that models with similar skills will soon become widely available, but did not give a specific timeframe for mass release.

This article was AI-translated and verified by a human editor