Unverified85% confidenceFactTime unknown
GPT 5.5在高难度合成对抗性攻击场景下，生物安全相关的拒绝率大约降低了一半
1
Sources
85%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026
Sources
GPT 5.5即时版深度解析：幻觉率腰斩背后的能力与安全博弈
bilibili英文白斑马
Related Entities
Related Claims
UnverifiedGPT-5.6的网络安全评测是在降低防护的条件下运行的，评测分数反映模型能力上限而非日常可用水平78% similarUnverifiedGPT 5.5即时版在医疗和法律领域的幻觉率大约降低了一半77% similarUnverifiedGPT-5.6在SWE-bench评测中以不到竞品一半的成本实现更优效果和更高解决率73% similarUnverifiedGPT-5.5在Databricks的Agent Harness设置下相比GPT-5.4实现了46%的错误率降低73% similarUnverifiedOpenAI tested GPT 5.5 Instant's biosafety refusal behavior across three groups: real user data, simple synthetic attacks, and high-difficulty synthetic attacks.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/3407API
curl https://kongchang.com/api/v1/knowledge/claims/3407MCP
get_claim(id=3407)