Anthropic released Claude Opus 4.8 this week, an upgrade to its most capable model that improves reliability across coding, research, legal, and financial workflows.

Here are the key updates:

• Better performance on complex professional work: Opus 4.8 improves on Opus 4.7 across benchmarks for coding, agentic skills, reasoning, and practical knowledge work.
• Effort control: You can now tell Claude how much thinking to put into a response. Higher effort means deeper reasoning; lower effort means faster replies that use fewer of your monthly credits.
• Improved honesty and judgment: Opus 4.8 is significantly less likely to gloss over mistakes or make unsupported claims about its own work. In agentic tasks like code review or document analysis, it flags uncertainties rather than pushing through with false confidence.
• Dynamic workflows in Claude Code: Claude can now plan a large task, spin up hundreds of parallel sub-agents to execute it, and verify the outputs before reporting back.
• Mid-task instruction updates: Developers can now update Claude's instructions partway through a long task without restarting the session. For teams building internal AI tools, this reduces friction in multi-step workflows that require context to shift as the task progresses.

Opus 4.8 is available now across all Claude plans at the same price as Opus 4.7. Effort control is available on all plans. Dynamic workflows in Claude Code are available on Enterprise, Team, and Max plans.

If your team uses Claude for research, drafting, analysis, or any workflow where Claude runs tasks on your behalf, this release gives you more control over quality, speed, and cost in a single session. It is also a sign that AI reliability is improving in ways that matter for real work.

​Explore Opus 4.8​

Read more AI news after this: