Unverified85% confidenceFactTime unknown

此前的HealthBench健康基准测试存在被游戏化的问题：模型给出的答案越长，得分就越高

1

Sources

85%

Confidence

Long-term

Relevance

5/31/2026

First Seen

Sources

GPT 5.5 Instant Deep Dive: The Capability vs. Safety Tradeoff Behind Halved Hallucination Rates

bilibili英文白斑马

Related Entities

HealthBenchproduct

Cite This Claim

Stable URI

https://kongchang.com/claim/3403

API

curl https://kongchang.com/api/v1/knowledge/claims/3403

MCP

get_claim(id=3403)