Anthropic released Claude Opus 5.5 on September 22, saying the AI model performs as well as its Claude Fable 5.1 system on most tasks while lowering operating costs compared with its predecessor, Claude Opus 5.

The company said Claude Opus 5.5 costs 40% less to run on typical workloads than Opus 5.

Anthropic priced the model at US$4 per million input tokens and US$20 per million output tokens, 20% below Opus 5.

Anthropic also said the model outperformed its recent AI models in internal safety tests. It said Claude Opus 5.5 was 85% less likely than Opus 5 or Mythos 5.1 to try to bypass its prescribed boundaries.

🔗 Source: Anthropic

Recent Anthropic developments