Claude 3.7 Sonnet and Claude Code Release

Anthropic has announced Claude 3.7 Sonnet, the first hybrid reasoning model that integrates both standard LLM capabilities and extended, step-by-step thinking into a single model. This release introduces a unified approach to reasoning, allowing users to choose between immediate responses and deep reflection to improve performance on complex technical tasks.

Claude 3.7 Sonnet: Hybrid Reasoning Capabilities

Claude 3.7 Sonnet functions as both a standard LLM and a reasoning model. In standard mode, it serves as an upgraded version of Claude 3.5 Sonnet. In extended thinking mode, the model self-reflects before answering, which enhances its performance in math, physics, coding, and instruction-following.

Key technical features of the reasoning system include:

  • Visible Thinking: The model's step-by-step reasoning process is made visible to the user.
  • Controllable Thinking Budgets: API users can specify a thinking budget (up to 128K tokens), allowing them to trade off response speed and cost for higher answer quality.
  • Real-World Optimization: Anthropic shifted focus from competition-style math and computer science problems toward real-world tasks that reflect actual business use cases.

Performance and Benchmarks

Claude 3.7 Sonnet achieves state-of-the-art performance on two primary benchmarks:

  • SWE-bench Verified: Evaluates the model's ability to solve real-world software issues.
  • TAU-bench: Tests AI agents on complex real-world tasks involving user and tool interactions.

Early testing from industry partners indicates high efficacy in coding. Cursor reported the model as best-in-class for real-world coding tasks and complex codebases. Cognition noted superior planning for code changes and full-stack updates. Vercel highlighted precision in complex agent workflows, and Replit used the model to build sophisticated web apps from scratch. Canva reported production-ready code with fewer errors and superior design taste.

Claude Code: Agentic Coding Tool

Alongside the model, Anthropic introduced Claude Code, a command-line tool currently in limited research preview. Claude Code acts as an active collaborator that can:

  • Search and read code.
  • Edit files and run tests.
  • Commit and push code to GitHub.
  • Use command-line tools.

Anthropic reports that in early testing, Claude Code completed tasks in a single pass that previously required over 45 minutes of manual work.

Availability and Pricing

Claude 3.7 Sonnet is available on all Claude plans (Free, Pro, Team, and Enterprise), the Claude Developer Platform, Amazon Bedrock, and Google Cloud's Vertex AI. Extended thinking mode is available on all surfaces except the free tier.

Pricing remains consistent with previous versions:

  • Input tokens: $3 per million tokens.
  • Output tokens: $15 per million tokens (including thinking tokens).

Safety and Reliability

Claude 3.7 Sonnet reduces unnecessary refusals by 45% compared to its predecessor, making more nuanced distinctions between harmful and benign requests. The release is accompanied by a system card detailing Responsible Scaling Policy evaluations and addressing risks associated with computer use, such as prompt injection attacks.

Sources

Related

  • Dispatch
    Claude 3.7 Sonnet Extended Thinking Release

    Anthropic has released Claude 3.7 Sonnet, featuring a toggleable extended thinking mode that allows the model to allocate more computational effort to complex problems and provides users with a visible raw thought process.

  • Dispatch
    Claude 3.5 Sonnet release notes / what's new

    Anthropic has released Claude 3.5 Sonnet, a model that outperforms Claude 3 Opus and competitors on reasoning and coding benchmarks while operating at twice the speed of Opus.

  • Dispatch
    Anthropic Claude Sonnet 4.5 release: state‑of‑the‑art coding model and new developer tools

    Anthropic announced Claude Sonnet 4.5, its most capable and aligned frontier model, delivering top‑tier coding, reasoning, and computer‑use performance while launching the Claude Agent SDK and major product upgrades.

  • Dispatch
    Claude 4 release notes / what's new

    Anthropic has released Claude Opus 4 and Claude Sonnet 4, introducing state-of-the-art coding capabilities, extended thinking with tool use, and the general availability of Claude Code.

  • Dispatch
    Anthropic "think" tool for Claude

    Anthropic introduces the "think" tool, a dedicated reasoning space that improves Claude's agentic tool use, policy adherence, and sequential decision-making in complex tasks.