Anthropic Introduces Claude Opus 5.5 With Lower Pricing

Anthropic has announced Claude Opus 5.5's release on September 22, 2026. As the first model in the new Claude 5.5 family, it performs at the level of Claude Fable 5.1 on most work while requiring 40% less compute to serve than Opus 5.

On knowledge work benchmarks like GDPval-AA v2.1, Opus 5.5 scores 1846, ahead of Claude Fable 5.1 at 1735 and Opus 5 at 1708. For agentic coding on Terminal-Bench 4.0, it records 66.4% at its highest effort level. On FrontierCode v1.1, the model scores 54.4%, and on CursorBench 4.0, it reaches 57.8%. In multidisciplinary reasoning on Humanity's Last Exam with tools, it achieves 67.7%, while scoring 81.8% on partial OSWorld 2.0 computer use tasks and 89.0% on visual chart recognition with tools.

Pricing for input and output tokens is set at $4 and $20 per million respectively, marking a 20% decrease from Opus 5. Cache reads cost $0.20 per million tokens, down 60% from Opus 5, and cache writes cost $5 per million tokens. A Fast mode is available in Claude Code and the Claude Platform providing up to 2.5x speed at $8 per million input tokens and $40 per million output tokens.

The company deployed the model with safeguards similar to Claude Fable 5.1 due to its comparability to Claude Mythos 5.1 in biology and cybersecurity. Vetted organizations can apply to the Life Sciences Verification Program to use the model for biology research, with cybersecurity verification access expanding in the coming weeks. Claude Sonnet 5.5 and Claude Haiku 5.5 are scheduled to follow in the coming weeks.