Unverified50% confidenceFactExact time
Taking GPT-MoE as an example, if the total number of experts is 64 and each token only activates Top-2, the actual computation can be reduced by approximately 97% compared to a Dense model of equivalent scale
1
Sources
50%
Confidence
Long-term
Relevance
10/6/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn 96% of levels, GPT-6 required fewer attempts to find the answer than the average person.76% similarUnverifiedGPT-6's claimed 50% price reduction refers to the API unit price, while Claude's claimed 40% reduction refers to task cost in official benchmarks; the two use different calculation methods and cannot be directly compared.70% similarUnverifiedThe gpt-4o-mini judge mislabeled as high as 77.1% of human cases on the enriched set that should have been judged as faithful70% similarUnverified一项针对GPT-4的研究显示，对MMLU题目进行语义等价改写后，模型准确率在部分子集上下降超过10个百分点69% similarUnverifiedIn MoE models, the weight distributions of different experts vary significantly, and aggressive quantization may cause disproportionate accuracy loss for certain experts.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/982627API
curl https://kongchang.com/api/v1/knowledge/claims/982627MCP
get_claim(id=982627)