
Summaries like this, in your inbox every morning.
Sign up free →Anthropic released Opus 4.8 on Thursday at standard pricing matching the previous Opus release, arriving 41 days after Opus 4.7 and significantly faster than the typical upgrade cycle for Sonnet and Haiku models (three and seven months old, respectively).
Opus 4.8 is designed to better flag uncertainties and avoid unsupported claims. Anthropic's testers found the model "more likely to flag uncertainties about its work and less likely to make unsupported claims." A testimonial from Bridgewater Associates highlighted the upgrade's tendency to "proactively flag issues with the inputs and outputs of an analysis."
Dynamic Workflows, available in research preview, allows Opus to manage complex tasks across hundreds of parallel subagents (AI agents that carry out work independently). Claude Code alongside Opus 4.8 can perform codebase-scale migrations across hundreds of thousands of lines of code from kickoff to merge.
Anthropic expects to bring Mythos-class models to all customers "in the coming weeks" after completing necessary safeguards, following a tentative preview last month that raised cybersecurity concerns.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started Free5 minutes a day. The AI essentials.
200+ sources · Email / LINE / Slack