Particle.news
Download on the App Store
15 ARTICLES
·
43m ago

Anthropic Launches Claude Sonnet 4.5 With 30-Hour Autonomous Runs and New Agent Tools

Anthropic pitches the model for business use, citing benchmark gains, agent tooling, plus new Microsoft 365 Copilot features.

Anthropic logo is seen in this illustration created on May 20, 2024. REUTERS/Dado Ruvic/Illustration/File Photo
Image
Claude 4

Overview

  • Anthropic released Claude Sonnet 4.5, available to all users and set as the default, and says it can operate autonomously for about 30 hours on complex, multi-step work.
  • Reported results show 77.2% on SWE-Bench Verified (up to 82% with parallel test-time compute) and 61.4% on OSWorld for real-world computer use.
  • The launch packages an agent-focused toolkit including a Claude Agent SDK, access to virtual machines and memory, improved context management, multi-agent support, and a native VS Code extension.
  • A limited “Imagine with Claude” research preview generates software on the fly for Max subscribers for five days, while Claude Code adds checkpoints and enhanced terminal workflows.
  • Microsoft said new Microsoft 365 Copilot capabilities will use Anthropic models, and Anthropic highlights safety gains such as reduced sycophancy and deception and stronger prompt-injection resistance.