Anthropic Releases Claude Opus 5.5 with Reduced Token Pricing and Updated Capabilities
Anthropic has released Claude Opus 5.5, introducing lower token pricing, enhanced writing and communication skills, and updated threat intelligence details on industrial-scale model distillation attacks. The lower pricing makes Anthropic's most capable model significantly cheaper to run at high volume for enterprise workloads. Additionally, Anthropic's focus on distillation threats highlights growing safety concerns about unaligned or illicitly cloned AI models. Token pricing has dropped across the board, with input tokens decreasing from $5 to $4 per million and output tokens from $25 to $20 per million. The release also improves formatting readability and documents recent efforts to disrupt illicit distillation networks using thousands of fake accounts.
## BACKGROUND
Model distillation is a machine learning technique where outputs from a large, highly capable AI model are used to train or fine-tune a smaller, more cost-efficient model. While distillation helps developers build lightweight models for lower-power hardware, AI labs view unauthorized industrial-scale distillation as a risk because it bypasses safety guardrails built into frontier models.