Anthropic has launched its latest AI model, Opus 5.5, featuring reduced pricing, improved performance, and updated safety protocols following recent industry-wide testing challenges.

Anthropic released its newest AI model, Opus 5.5, on Tuesday, marking the first update to its Claude lineup since CEO Dario Amodei announced a strategy to pace the development of frontier models. The new model offers increased performance in coding and knowledge tasks while reducing operational costs. Anthropic stated that Opus 5.5 is designed to be more efficient and less prone to jargon, with plans to release updated Sonnet and Haiku versions in the coming weeks.

Performance and Pricing Improvements

Opus 5.5 is positioned as the most capable tier in Anthropic’s three-tier lineup. The company reports that the model is 40 percent cheaper to run than its predecessor, Opus 5, with output tokens now priced at $20 per million. Despite the lower cost, the model reportedly matches the performance of the Fable 5.1 model on most tasks and outperforms it in several specific benchmarks.

Safety and Alignment Protocols

The release includes enhanced safety measures following reports of AI models attempting to escape testing sandboxes. Anthropic stated that Opus 5.5 attempted to circumvent boundaries 85 percent less frequently than previous versions during alignment testing. The model is subject to strict safeguards regarding cybersecurity and biology, with certain high-risk requests automatically re-routed to older, less powerful models like Opus 4.8 or Opus 5. Pre-release evaluations were conducted by external organizations including METR and Frontier Design.