Anthropic has released Claude Opus 4.8, the latest version of its flagship AI model, making it available to users and developers globally at the same price as its predecessor. This update improves coding, reasoning, agentic skills, and practical knowledge work, with benchmarks showing that Opus 4.8 outperforms previous Opus models and even matches or surpasses leading competitors on key tasks. Fast mode now delivers responses at 2.5 times the previous speed at a third of the former cost.
New features include effort control for claude.ai, allowing users to decide how deeply the model processes tasks, and dynamic workflows in Claude Code, enabling parallel handling of very large-scale problems. Opus 4.8 is accessible on all existing platforms, including the Claude website and API, for both individual and enterprise users. The new dynamic workflows feature is in research preview for Enterprise, Team, and Max plans, while effort control is available to all users. Early users and industry partners report that Opus 4.8’s judgment and reliability are improved, especially in agentic and legal applications, and that it performs more rigorous self-checks, reducing unflagged errors.
The model’s alignment and safety have also undergone extensive assessment, confirming lower rates of misaligned behavior compared to earlier versions. Anthropic continues to position itself at the forefront of responsible AI development, with plans to bring even more advanced models to market in the near future.