Unverified50% confidenceBenchmarkExact time

Luna配置在HIGH和XH模式下10次测试全部答错，正确率为零

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/20/2026

First Seen

Valid until: 10/18/2026

Sources

GPT-5.6 Codex三模型实测：Sol/Terra/Luna糖果逻辑题横评

Related Claims

UnverifiedSol配置在HIGH模式跑10次前6次全部答对，XH模式多次运行均保持全对成绩62% similar Unverified在盲测中Sol给Opus方案打8.3分、给Fable方案打6.0分56% similar Unverified早期在Opus模型上测试精确字符串识别率为15分之0，而Fable 5发布后同样测试变成15分之1354% similar UnverifiedQwen 3.6 Max中性模式下歧义检测F1仅16%，平均每任务只问0.07个问题，但一旦提问澄清准确率高达94.7%54% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/573188

API

curl https://kongchang.com/api/v1/knowledge/claims/573188

MCP

get_claim(id=573188)