Event ·Single source

测试显示 GPT-6.1 Sol 在研究任务上引用造假率为 0%

一项针对引用造假（citation hallucination）的测试发现，新一代模型的引用捏造比例明显低于旧版模型。其中 GPT-6.1 Sol 在研究任务上造假引用的比例为 0%。长上下文中的刻板印象与事实丢失问题也有改善，但远称不上完美。

Current read

有博主测了各代模型编造参考文献的情况，GPT-6.1 Sol 造假引用降到 0%，长上下文丢事实也有改善，做文献调研的可以看看。

2 sources29° AI heatLast updated 2026/11/3 12:27:02

Evidence

2 sources

Primary source@koltregaskes

2026/11/3 12:27:02View source

Related source 1elvis

2026/11/3 18:29:49View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story