Particle.news
Download on the App Store
32 ARTICLES
·
2h ago

Anthropic Launches Claude Sonnet 4.5 With Benchmark Gains and 30-Hour Autonomy

Priced the same as Sonnet 4, the release targets enterprise coding use cases through new SDKs, memory tools, plus a VS Code extension.

Anthropic logo is seen in this illustration created on May 20, 2024. REUTERS/Dado Ruvic/Illustration/File Photo
Abstract composition with flowing head silhouette and stylized hand gesture, holding geometric cube structure
Image

Overview

  • Anthropic says Sonnet 4.5 leads key tests, scoring 77.2% on SWE‑Bench Verified and 61.4% on OSWorld, and calls it its best model for coding and computer use.
  • The model reportedly maintained focus for about 30 hours on complex, multi‑step tasks, a jump from seven hours on Opus 4.
  • Sonnet 4.5 is available now in the Claude chatbot and API at $3 per million input tokens and $15 per million output tokens, with Anthropic indicating it will be the default experience for users.
  • Companion updates include Claude Code checkpoints, a refreshed terminal, a native VS Code extension, a Claude Agent SDK, context editing and a memory tool, plus a five‑day ‘Imagine with Claude’ preview for Max subscribers.
  • Anthropic touts improved alignment and ASL‑3 protections, though early third‑party tests reported a jailbreak, underscoring ongoing safety and verification work.