Verified80% confidenceFactExact time
OpenAI表示投入超过70万小时A100等效算力用于自动化红队测试
6
Sources
80%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
Valid until: 10/7/2026
Sources
GPT-5.6三模型详解：Sol、Terra、Luna分层定位与能力对比
bilibiliX信息差6/28/2026
Related Claims
UnverifiedOpenAI投入超过70万GPU小时用自己的模型主动挖掘通用越狱漏洞79% similarVerifiedOpenAI的o1系列代表了「推理时计算」（test-time compute）这一新范式71% similarUnverifiedOpenClaw开发团队加入了每6小时对NPM发布进行自动化测试的机制68% similarPartially VerifiedOpenAI内部曾用三个人、五到六个月时间，通过Harness Engineering方式生成了超过一百万行代码并成功上线68% similarUnverifiedOpenAI的Evals框架和PromptFlow提供了Prompt的自动化测试能力67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/374271API
curl https://kongchang.com/api/v1/knowledge/claims/374271MCP
get_claim(id=374271)