Particle.news
Download on the App Store
38 ARTICLES
·
last wk.

Anthropic Unveils Claude Sonnet 4.5 With 30-Hour Autonomy and Top Coding Scores

Anthropic pairs benchmark gains with new agent tools at the same price.

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

  • The new model sustained roughly 30 hours of autonomous, multi‑step software work, up from about seven hours reported for Opus 4.
  • Anthropic reports state‑of‑the‑art results on real‑world tests, scoring 77.2% on SWE‑Bench Verified (82% with parallel compute) and 61.4% on OSWorld.
  • Developer rollouts include Claude Code checkpoints, a native VS Code extension, a refreshed terminal, a Claude Agent SDK, and API features for memory and automatic context management.
  • Pricing remains unchanged from Sonnet 4 at $3 per million input tokens and $15 per million output tokens across the API and tools.
  • Anthropic touts ASL‑3 safety measures and reduced susceptibility to prompt injection, though independent testers reported successful jailbreaks shortly after release.