Anthropic launches Claude Opus 4.8 with faster, cheaper mode

Anthropic has just rolled out Claude Opus 4.8, touting sharper judgment, more honesty, and longer independent work. A new “Fast mode” promises roughly 2.5x speed and lower cost, while Claude Code gets deeper agentic workflows for long-running tasks.

claude cover

TL;DR

  • Claude Opus 4.8: Update to Opus 4.7, with “sharper judgment” and “more honesty about its own progress”
  • Longer independent work: Claimed ability to operate without constant check-ins and stay on track in long sessions
  • Available today: Same price as before
  • Fast mode: Same model at ~2.5x speed and “three times cheaper than before”
  • Claude Code trigger: Fast mode enabled via /fast; API access limited via account managers or waitlist
  • Dynamic workflows (research preview): Plans, runs hundreds of parallel subagents, verifies work; example: large migrations across hundreds of files

Anthropic’s Claude account posted a launch thread on X for Claude Opus 4.8, describing the new model as an update to Opus 4.7 with “sharper judgment,” “more honesty about its own progress,” and “the ability to work independently for longer than its predecessors.” The company also claimed the model is “available today at the same price.”

The thread also introduced a “Fast mode” for Opus 4.8. Anthropic says it is “the same model at roughly 2.5x the speed,” while also being “three times cheaper than before.” The option is enabled with /fast in Claude Code, and API access appears to be limited for now, with the company directing customers to account managers or a waitlist. A separate post linked from the thread carries the fast-mode details.

Claude Code is getting a more prominent role in the release. Anthropic claims Opus 4.8 “makes calls like an experienced engineer without needing constant check-ins,” stays on track in long-running sessions, and can follow work through a repo so tasks can be handed off while other work continues. The company also pointed to “dynamic workflows” in Claude Code as a research preview. In that setup, Claude “makes a plan, runs hundreds of parallel subagents, and verifies its work before reporting back,” with migrations touching hundreds of files offered as an example. The company’s read-more link points to the feature.

Source: Claude on X

Continue the conversation on Slack

Did this article spark your interest? Join our community of experts and enthusiasts to dive deeper, ask questions, and share your ideas.

Join our community