Anthropic Launches Claude 3.7 Sonnet with Integrated Reasoning, Debuts Developer Tool
Amazon-backed AI company Anthropic has unveiled its latest language model, Claude 3.7 Sonnet, entering the increasingly competitive race against rivals like OpenAI's ChatGPT and China's DeepSeek. In a significant development, the company has also introduced Claude Code, its first venture into AI agents for developers.
Unlike traditional approaches that separate pre-trained and reasoning models, Anthropic has integrated reasoning capabilities directly into its frontier model. This innovation allows users to activate an "extended thinking mode" when they want the AI to think longer before responding to complex queries. The company claims this feature significantly improves performance in mathematics, physics, coding, and instruction following.
Reasoning models, designed to mimic human-like thinking for solving complex problems, have become a significant focus on the AI industry since OpenAI released its first reasoning model, o1, in September last year. Since then, Google, DeepSeek, and Elon Musk's Grok have all launched similar capabilities.
Anthropic emphasizes that its reasoning model has been optimized for real-world applications rather than focusing narrowly on mathematics and computer science problems. The company has released benchmark data showing Claude 3.7 Sonnet outperforming competitors like OpenAI's o-3 Mini (High) and DeepSeek R1 in accuracy tests.
However, while competitors like Grok 3 Reasoning, DeepSeek R1, and o3 Mini offer free access (with bandwidth limitations), Anthropic's extended reasoning mode will be exclusive to paying subscribers. Free users can still access the standard Claude 3.7 Sonnet model without the extended reasoning feature.
Anthropic's new Claude Code represents the company's first agentic tool offering. Currently available in limited preview, this command-line tool enables developers to delegate complex engineering tasks directly from their terminal. The company has also improved the coding experience on its website, and its GitHub extension is available across all subscription tiers, allowing developers to connect coding repositories directly to the chatbot.
According to Anthropic, Claude Code can "search and read code, edit files, write and run tests, commit and push code to GitHub, and use command-line tools" while maintaining user oversight.
The announcement comes amid intensifying competition in the AI industry, particularly following Chinese AI startup DeepSeek's January 2025 release of its cost-efficient DeepSeek-R1 model. DeepSeek's claim of training a ChatGPT-rivaling model for just $6 million—compared to the billions spent by U.S. competitors—sent shockwaves through Silicon Valley, triggering a market response that saw Nvidia lose nearly $600 billion in value in a single day.