Anthropic PBC has released Claude Opus 4.8, a new large language model that delivers significant improvements in complex coding and reasoning tasks.
The company also announced a massive $65 billion funding round at a $965 billion valuation, making it worth more than OpenAI. The capital will be used to expand computing infrastructure.
On Terminal-Bench 2.1, Claude Opus 4.8 scored 74.2%, an 8.4% gain over its predecessor. It also posted a 4.9% improvement on the SWE-Bench Pro coding test. The model is four times less likely to output faulty code without flagging the error.
A new feature, dynamic workflows in Claude Code, allows the assistant to rewrite entire applications by spawning hundreds of specialized AI agents. An upgraded fast mode boosts speed by 150% while cutting costs by two-thirds.
Looking ahead, Anthropic plans to release Mythos-class models broadly in the coming weeks. Claude Mythos Preview, already deployed to select organizations, excels at cybersecurity vulnerability detection. The company is developing guardrails to safely expand access.
Anthropic’s run rate has tripled to $47 billion in three months. The Series H round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, with co-leads including Capital Group, Coatue, and GIC. AWS committed $5 billion.