Anthropic has released Claude Opus 5.5, the first model in its new Claude 5.5 family. This release is notable for achieving performance comparable to the Claude Fable 5.1 while reducing running costs by 40% compared to its predecessor, Opus 5.
Claude Opus 5.5 is designed to excel in agentic coding, computer use, and knowledge work, according to Anthropic's benchmarks. It is available as a managed API model and can be accessed via the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure. The model does not allow for self-hosting as weights have not been released. As with previous Opus models, zero data retention is an option.
The model demonstrates strong performance on several benchmarks, such as achieving 66.4% on Terminal-Bench 4.0 and 54.4% on FrontierCode v1.1, surpassing both Claude Fable 5.1 and Opus 5. Cost-adjusted results highlight its efficiency, with a score of 54.6% on FrontierCode at default effort, outperforming GPT-6 Astra at significantly lower costs. Pricing for Claude Opus 5.5 reflects its reduced computational needs, with input costs at $4 per million tokens and output costs at $20 per million tokens, which are lower than those for Opus 5.
Early testers have reported significant improvements in efficiency and speed, such as completing large code migrations and audits much faster than with previous models. Notably, Opus 5.5 also demonstrated improved accuracy in quality checks and bug detection.
Claude Opus 5.5 is targeted at developers and enterprises looking for efficient and cost-effective AI solutions. It is suitable for tasks involving code migration, auditing, and complex knowledge work, providing a robust tool for software development and enterprise automation.
Work implications: Claude Opus 5.5 can enhance workflows in software development and enterprise operations by speeding up code migrations and audits, potentially reducing the time and cost associated with these tasks.
Originally reported by Marktechpost AI Media