Anthropic Releases Claude Opus 5.5 With Lower Prices and Stronger Safeguards

Summary

Anthropic has released Claude Opus 5.5 across its own platform and major cloud providers, targeting long-running enterprise tasks. The model is faster and less expensive than Opus 5, and includes preserved-thinking protections, watermarking, and stronger defenses against prompt injection. Anthropic also introduced controls for sensitive work, including screening coding-agent actions and routing most cybersecurity tasks to Opus 4.8.

Key points

• Claude Opus 5.5 is available through Anthropic, Amazon Web Services, Google Cloud, and Microsoft Azure.
• The model is designed for lengthy tasks such as code migrations, software audits, financial analysis, and multi-application workflows.
• Anthropic reports faster output and lower token use than Opus 5 in tests; results vary by task and settings.
• Preserved thinking is intended to prevent API users from editing prior context and make large-scale model distillation harder. It applies to API accounts created on or after 2026-08-31.
• New safeguards include screening coding-agent actions, vulnerability-checking features, and stronger resistance to prompt-injection attempts. Most cybersecurity tasks are routed to Opus 4.8.
• API prices are lower than Opus 5, output is more than 30% faster, and Anthropic is increasing some subscription usage limits.

Tags

Vendors

Products

Industries