Anthropic has launched Claude Opus 5.5, marking the debut of its Claude 5.5 model family. Available immediately as a managed API across the Claude Platform, AWS, Google Cloud, and Microsoft Azure, the model matches or exceeds Claude Fable 5.1 capabilities across agentic coding, computer use, and complex knowledge work, while reducing typical operating costs by 40% compared to Opus 5.

Opus 5.5 achieves its cost efficiency through reduced compute requirements, faster output generation, and a 60% price drop on cache reads. On evaluation benchmarks, medium-effort Opus 5.5 achieved a 54.6% score on FrontierCode, outperforming rival models at a fraction of the cost per task. Anthropic also expanded subscriber usage limits and introduced a fast mode delivering up to 2.5x operational speeds.

Safety and regulatory compliance features feature prominently in this release. Opus 5.5 incorporates mandatory thinking processes, automated behavioral safeguards for biology and cyber risks, and embedded output watermarking to comply with the EU AI Act. Evaluators noted an 85% drop in boundary circumvention attempts compared to its predecessor.

Why it matters

  • Enterprise API users can lower agentic coding and knowledge work expenses by 40% while gaining faster execution times.

  • Built-in watermarking directly addresses European compliance, easing regulatory overhead for deployments in the EU market.

  • Non-optional thinking steps and strict safety routing reflect a permanent shift toward enforceability in high-risk frontier models.

Source: marktechpost.com