Anthropic unveils Claude 3.7 Sonnet and Claude Code
Created on February 25|Last edited on February 25
Comment
Claude 3.7 Sonnet represents the latest leap in AI intelligence from Anthropic, introducing hybrid reasoning capabilities that allow for both near-instant responses and deeper, step-by-step thinking. This makes it the first model of its kind, providing users with control over how much time the AI spends on reasoning. API users can set limits on the number of tokens allocated to extended thinking, balancing speed and accuracy to fit different use cases.
Improved Coding and Software Development Capabilities
One of the most significant improvements in Claude 3.7 Sonnet is its ability to handle complex coding tasks, particularly in front-end web development. Alongside this, Anthropic has launched Claude Code, a new command-line tool designed to assist developers by automating substantial engineering tasks directly from the terminal. Claude Code is currently in limited research preview but has already demonstrated impressive performance in test-driven development, debugging, and large-scale code refactoring.
Availability and Pricing
Claude 3.7 Sonnet is available across all Claude plans, including Free, Pro, Team, and Enterprise, as well as through the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI. However, extended thinking mode is not included in the free tier. Pricing remains the same as previous versions, at $3 per million input tokens and $15 per million output tokens, including thinking tokens.
A New Approach to AI Reasoning
Anthropic has developed Claude 3.7 Sonnet with a philosophy that integrates quick responses and deep reasoning within the same model. Unlike other AI systems that separate these functions, Claude 3.7 Sonnet allows users to choose when they need a fast answer versus a more thoughtful, multi-step response. Extended thinking mode particularly enhances performance in math, physics, coding, and instruction-following tasks, making it well-suited for real-world applications rather than just competition-based benchmarks.
Performance in Coding and AI Benchmarks
Claude 3.7 Sonnet has set new standards in software engineering AI, outperforming previous models in real-world coding tasks. Independent evaluations from organizations like Cursor, Cognition, Vercel, and Replit confirm its superiority in handling complex codebases, full-stack updates, and advanced tool use. Canva’s testing found that it consistently generated production-ready code with better design choices and fewer errors. The model also achieved state-of-the-art results on SWE-bench Verified and TAU-bench, two frameworks that measure AI capabilities in solving software issues and executing real-world tasks.
Claude Code: A New Tool for AI-Driven Software Development
Anthropic’s introduction of Claude Code expands the AI’s capabilities beyond just responding to prompts. This agentic coding tool can read and modify code, edit files, run tests, commit changes to GitHub, and use command-line tools while keeping the developer informed. In early testing, Claude Code significantly reduced the time required for tasks like debugging and refactoring, often completing work in a single pass that would otherwise take over 45 minutes. Future updates will focus on improving tool reliability, supporting long-running commands, and enhancing in-app rendering.
GitHub Integration and AI-Powered Collaboration
Beyond Claude Code, Anthropic has enhanced the overall coding experience by integrating Claude with GitHub across all plans. This allows developers to link their repositories directly to Claude, making it an even more effective tool for fixing bugs, developing features, and generating documentation.
Looking Ahead
The release of Claude 3.7 Sonnet and Claude Code marks a major step toward AI systems that can more effectively augment human abilities. With deep reasoning capabilities, enhanced autonomy, and improved collaboration tools, these advancements bring AI closer to becoming an indispensable partner in coding, problem-solving, and creative work. As Anthropic continues to refine its models, user feedback will play a crucial role in shaping the future of AI development.
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.