Claude Opus 5.5 matches Fable 5.1 performance at lower cost and promises less "Claudish" writing
decoder ·
Anthropic has launched Claude Opus 5.5, a new AI model that matches the performance of Fable 5.1 while reducing operational costs by approximately 40 percent. The update features lower token prices and faster generation speeds, positioning it competitively against rivals like OpenAI's GPT-6 Astra in benchmarks for agentic coding and knowledge work. A key improvement addresses previous criticisms regarding the model's formulaic writing style, with Anthropic promising more natural and concise communication to reduce what users called "Claudish" output. The company also introduced stricter safety safeguards for cybersecurity and biological research, routing flagged requests to specialized models rather than the main instance. To combat distillation attacks, Opus 5.5 includes anti-extraction measures and watermarking to ensure compliance with regulations such as the EU AI Act. Additionally, Anthropic is calling for higher international safety standards as models become capable of automating their own research. This release marks the beginning of a new family, with Sonnet and Haiku variants expected in the coming weeks.
decoder