Unverified50% confidenceFactExact time
当Cursor检测到账号使用异常时，可能会将请求路由到性能较低的模型版本或降低响应质量，这被称为模型降智
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/28/2026
First Seen
Valid until: 11/26/2026
Sources
Related Entities
Related Claims
Unverified不同模型在基准测试上的性能差异往往远小于同一模型在不同任务类型、不同输入格式下的行为差异69% similarUnverifiedResearch shows that a model's accuracy on self-criticism tasks is significantly lower than in scenarios where criticism is performed by another model.69% similarUnverifiedThe intermediary layer can wrap responses from a lower-version model as output from the latest version, and ordinary users can hardly distinguish which model version was actually called68% similarUnverifiedModels with more parameters generally have stronger comprehension capabilities, but respond more slowly and consume more tokens.68% similarUnverifiedUsers only need to lightly paraphrase AI output for detection scores to drop significantly.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/810574API
curl https://kongchang.com/api/v1/knowledge/claims/810574MCP
get_claim(id=810574)