Anthropic Unveils Latest AI Model ‘Claude Sonnet 4.5’...Competition with OpenAI Heats Up

U.S. AI startup Anthropic has released its next-generation artificial intelligence model, Claude Sonnet 4.5, further intensifying its rivalry with OpenAI. This launch follows the May release of Claude Sonnet 4 and the August debut of the premium model Claude Opus 4.1.

Anthropic emphasized that Sonnet 4.5 goes beyond being a simple conversational chatbot, serving instead as a colleague-like AI capable of assisting in specialized fields such as coding, business, research, and finance. Notably, it received recognition as “the world’s best coding model” after achieving top performance on the SWE-bench Verified benchmark for software engineering.

Performance and Safety Upgrades

The most distinctive feature of the new model is its continuous coding ability. With just a single instruction, it can autonomously write and refine code for up to 30 hours—significantly surpassing the 7-hour limit of Claude Opus 4, released in May.

The model also boasts enhanced performance in high-stakes domains including cybersecurity, finance, and scientific research. According to Anthropic, it underwent large-scale safety training to reduce misleading or preference-based responses, ensuring more trustworthy outputs.

“Smaller but Smarter” Than Opus 4.1

Jared Kaplan, Anthropic’s co-founder and Chief Science Officer (CSO), told CNBC:

“Claude Sonnet 4.5 doesn’t just feel smarter—it feels like a teammate that helps you solve problems.”

He added that although the model is smaller in size compared to Opus 4.1, it outperforms it in nearly every aspect. This reflects Anthropic’s dual achievement of model efficiency and performance optimization.

AI Market Showdown Expected Through Year-End

Alongside the Sonnet 4.5 announcement, Anthropic revealed plans to release additional models before the year’s end. Kaplan hinted that “one or two more model launches are expected,” with a strong likelihood of a new version of Opus.

Founded in 2021 by former OpenAI employees, Anthropic has grown to a valuation of around $183 billion (₩256 trillion), backed by major investments from Amazon and Google. Today, more than 300,000 companies worldwide use Claude, solidifying its position as a direct challenger to OpenAI’s ChatGPT.

By Jubaek Shin, The Korea Metaverse Journal

jbshin@kmjournal.net