Introducing Claude Opus 4.8
Anthropic released Claude Opus 4.8, claiming the top spot for computer-use and browser-agent tasks with an 84% score on Online-Mind2Web — a meaningful jump over Opus 4.7 and GPT-5.5. The release adds dynamic workflows in Claude Code for large-scale parallel tasks, user-controlled effort levels on claude.ai, and a fast mode running at 2.5× speed at the same price of $5/$25 per million input/output tokens.
anthropic.com ↗