Anthropic Launches Claude 2 Chatbot with Improved Coding and Reasoning Skills
Anthropic introduces Claude 2, an upgraded AI chatbot that shows significant improvements in coding, math, and reasoning over its predecessor. The new model is currently available in beta across the U.S. and U.K. for web users and select API developers.
Anthropic releases Claude 2, the second generation of its commercial AI chatbot, making it available in beta in the U.S. and U.K. The new model operates through a web interface and a paid API, maintaining the same pricing structure as the previous version while already attracting pilot programs from businesses like Jasper and Sourcegraph.
The upgraded chatbot outperforms its predecessor across multiple academic and technical benchmarks. Claude 2 achieves higher scores on the bar exam, passes the multiple choice section of the U.S. Medical Licensing Exam, and demonstrates significantly stronger programming abilities on the Codex Human Level Python coding test. The model also answers more grade-school-level math problems correctly, thanks to enhanced reasoning and better self-awareness regarding its limitations.
Anthropic trains Claude 2 on more recent data from early 2023, including a diverse mix of websites, licensed third-party datasets, and some non-English text. Despite these meaningful performance gains, the company describes Claude 2 as a continuously iterated, tweaked version of the Claude 1.3 architecture rather than a completely new architectural creation.