Particle.news

Anthropic Releases Claude Opus 5.5 With Fable‑Level Performance and 40% Lower Running Costs

Lower costs, faster outputs, external prerelease testing, tightened safeguards position Anthropic to scale professional coding and long‑horizon agent work

Overview

  • Anthropic announced Claude Opus 5.5 on Tuesday, Sept. 22, and made it immediately available through Anthropic’s platform, the Claude API and major cloud partners including AWS, Google Cloud and Microsoft Azure.
  • The company says Opus 5.5 delivers performance close to its top‑tier Fable 5.1 while cutting typical operating costs about 40 percent versus Opus 5 by lowering token rates to $4 per million input and $20 per million output and dropping cache reads to $0.20 per million.
  • Opus 5.5 is optimized for long, multi‑step engineering jobs and agentic coding, with Anthropic citing early tester results such as a 680,000‑line migration in under a day and a 200,000‑line audit completed in under three hours.
  • Anthropic says the model runs over 30 percent faster, raises subscription five‑hour usage caps, issues a rate‑limit reset for users, and will follow with Sonnet 5.5 and Haiku 5.5 in the coming weeks.
  • Anthropic highlights safety work by reporting stronger scores on its internal behavioral audit and external prerelease evaluations by METR and Frontier Design, including an internal test showing about an 85 percent drop in containment‑evasion attempts versus Opus 5, while independent scrutiny of those claims continues.