Introducing Claude Opus 4.8

Summary

Anthropic introduces Claude Opus 4.8, an upgrade over Opus 4.7 for coding, agentic skills, reasoning, and practical knowledge work, with regular pricing unchanged. The release adds adjustable effort in claude.ai and Cowork, dynamic workflows in Claude Code, fast mode at 2.5× speed, and support for system entries inside the Messages API messages array. Dynamic workflows can plan work, run hundreds of parallel subagents in one session, and verify outputs; the post cites codebase-scale migrations across hundreds of thousands of lines as an example. Early evaluations and tester reports describe stronger reliability, judgment, tool efficiency, computer use, and analysis, including an approximately fourfold reduction in allowing flaws in generated code to pass unremarked.

Takeaways

1. 1

   Opus 4.8 defaults to high effort, while users can select extra or max; higher settings spend more tokens, and lower settings respond faster while using rate limits more slowly.

2. 2

   Dynamic workflows are available in research preview for Claude Code Enterprise, Team, and Max plans, and can run codebase-scale migrations against an existing test suite.

3. 3

   Alignment evaluations found higher measures of supporting user autonomy and acting in the user’s best interest, with substantially lower misaligned behavior than Opus 4.7.