Anthropic's Claude 3.5 Sonnet outperforms OpenAI's GPT-4o in reasoning and coding metrics

From Observer: 2024-06-21 14:58:42

Anthropic, founded by former OpenAI engineers, released Claude 3.5 Sonnet, their most powerful A.I. model yet, surpassing GPT-4o in reasoning and coding metrics. The startup aims to release more models this year to enhance intelligence, speed, and cost tradeoffs. Anthropic’s 375 employees signify a push to compete with OpenAI’s 2,000 employees in the A.I. industry.

Claude outperforms GPT-4o in reasoning, coding, and math proficiency benchmarks. The new model aims to personalize user experiences with memory-focused capabilities. Despite impressive results, A.I. benchmarks may not accurately represent everyday interactions. Claude also excels in visual comprehension, essential for industries like retail and finance.

Anthropic’s new model includes a feature called Artifacts, creating a collaborative workspace for users to interact with generated content. Both versions of Claude 3.5 Sonnet are available for free, with paid subscriptions offering higher usage limits. Safety is a priority for Anthropic, conducting extensive safety tests before deployment.

Anthropic’s commitment to safety contrasts with recent criticisms of OpenAI’s safety protocols. Former OpenAI employee Jan Leike left to join Anthropic, emphasizing the importance of developing reliable and aligned A.I. systems. Dario Amodei expressed dedication to responsibly tackling the complex challenges of creating advanced A.I. models.



Read more at Observer: Anthropic Claude 3.5 Sonnet vs. OpenAI GPT-4o: Which Is Better?