基准测试

Anthropic 被指用不实信息给 新模型造势 — AI 公司的 「发布会公关」正在透支可信度

Analysts flag selective benchmark reporting in Anthropic's Claude Mythos launch—part of a wider AI industry credibility problem .