Unverified50% confidenceFactExact time

高考作文评分由GPT-4.1和Claude Sonnet 5两个模型交叉打分，模仿高考双评卷模式

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/9/2026

First Seen

Valid until: 10/7/2026

Sources

华为盘古2.0 Flash高考实测：613分成绩与稳定性深度解析

bilibiliwycode_cn7/7/2026

Related Claims

VerifiedClaude Sonnet 4和GPT-5被归类为高级模型，适合处理复杂推理任务和架构设计74% similar Unverified主流商业模型如GPT-4o、Claude 3.5 Sonnet按输入与输出token分别计费72% similar UnverifiedLLM-as-Judge范式使用能力更强的大语言模型（如GPT-4o或Claude 3.5 Sonnet）对被评估模型的输出进行打分72% similar Unverified评测采用 GPT-5.5 担任 LLM 评审法官，配合人工品味评分的混合方式71% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/330509

API

curl https://kongchang.com/api/v1/knowledge/claims/330509

MCP

get_claim(id=330509)