Anthropic yesterday (22 May) released Claude Opus 4 and Claude Sonnet 4 – models it claims set new standards for coding, advanced reasoning and AI.
Claude Sonnet 4 replaces Claude Sonnet 3.7, with better coding and reasoning while responding more precisely to instructions, claims Anthropic, which described Claude Opus 4 as its most powerful model yet and “the best coding model in the world”.
“Built for sustained performance on complex, long-running tasks, it can maintain focused effort across thousands of steps,” the company said of Opus 4.
Both of the new models are hybrid and offer two modes – near-instant responses or extended thinking for deeper reasoning.
“These models advance our customers’ AI strategies across the board,” Anthropic said in its statement. “Opus 4 pushes boundaries in coding, research, writing and scientific discovery, while Sonnet 4 brings frontier performance to everyday use cases as an instant upgrade from Sonnet 3.7.”
Anthropic, which recently announced a major European expansion including jobs in Dublin, also launched a whole slew of updates to Claude.
“Claude 4 models raise the bar for experienced engineers, making it easier to write, edit and debug code while powering the tools developers love – and open the door for complete beginners, from solopreneurs to sales leaders, putting creative power in the hands of everyone who wants to build,” it claims.
Anthropic was founded by siblings and Open AI alumni Dario and Daniela Amodei. Still a private company, it was built on the principles of responsible AI, although many commentators have questioned the ability of any AI player to be truly ethical.
In recent years it has received funding from the likes of Amazon and Google, drawing ire from online purists.
Anthropic released the new models with a slew of reviews from illustrious users gushing about their efficacy, but one caught my eye, a review of Claude Sonnet 4 from Tao Zhang, co-founder of the major Chinese AI player, Manus.
Manus launched a $39/month subscription offering in March, putting it in direct competition with the likes of OpenAI’s ChatGPT, and of course Claude – although Manus itself admits it is still in testing phase. Reported investors in Manus include Benchmark, Tencent Holdings, HSG (formerly Sequoia) and ZhenFund.
“Claude Sonnet 4’s ability to follow complex, multi-step instructions and work through problems with clear chain-of-thought reasoning is remarkable. The aesthetics of the artifacts are really excellent – I’ve never seen anything like it,” Zhang said.
As for Claude Opus 4, Yusuke Kaji, general manager, AI, at Rakuten, was mightily impressed: “Opus 4 offers truly advanced reasoning for coding,” he said. “When our team deployed Opus 4 on a complex open source project, it coded autonomously for nearly seven hours – a huge leap in AI capabilities that left the team amazed.”
High praise indeed. As a user of Claude myself – I have no illusions that Anthropic is perfect but I have great time for its ethical aspirations – I’m excited to get cracking on trying it out. Not for reporting, I hasten to add – we have a strict AI policy here at SiliconRepublic.com when it comes to journalism. I’ll report back once I’ve run the new model through its paces for more mundane tasks.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.