Leading AI company Anthropic has unveiled its latest model, Claude 3.7 Sonnet, along with a new tool for developers called Claude Code. Claude 3.7 Sonnet features “hybrid reasoning” capabilities, allowing it to provide rapid responses as well as step-by-step solutions to complex problems. This enhancement significantly improves performance in areas such as mathematics, physics, and coding. Users can adjust the model’s thinking duration and depth according to their specific needs. Claude 3.7 Sonnet is now accessible through the Claude app and is available to developers via Anthropic’s API, Amazon Bedrock, and Google Cloud’s Vertex AI platforms, all while maintaining the same cost structure as its predecessor. In addition, Anthropic has introduced Claude Code, a tool designed to assist developers in tasks such as code reading, editing, testing, and interacting with command-line tools, aiming to enhance collaborative coding experiences. The release of Claude 3.7 Sonnet marks a significant step towards the versatile application of AI systems.
Meanwhile, OpenAI’s o1 model has demonstrated superior performance in solving complex tasks and coding, bringing AI closer to human-like reasoning capabilities. However, benchmark tests reveal that Claude 3.7 Sonnet surpasses OpenAI’s o1 model, achieving higher scores in coding and agentic tool usage. Notably, Claude 3.7 Sonnet excelled in SWE-bench Verified tests, which assess the ability to resolve real-world software issues. Additionally, it showcased outstanding performance in TAU-bench evaluations, focusing on complex tasks involving user and tool interactions. These findings position Claude 3.7 Sonnet as a leader in coding and agentic tool applications.