Unverified50% confidenceBenchmarkExact time
According to official benchmark data from OpenViking, the reduction in input tokens ranges from 34% to 91%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/18/2026
First Seen
Valid until: 12/17/2026
Sources
Related Entities
Related Claims
UnverifiedUsing a prompt that simply says 'write less code' can reduce code volume by 33%, but the security metric drops to 95%.68% similarUnverifiedInference costs have dropped sharply, with the token cost of some models falling by more than 90%.66% similarUnverifiedAfter removing the candidate filtering component, FEVER accuracy dropped to 76.24%.66% similarUnverifiedWhen prefix caching and DFlash2 speculative decoding are both enabled simultaneously, total throughput drops by approximately 37%, which is even slower than not enabling speculative decoding at all.66% similarUnverifiedNielsen Norman Group研究表明用户遭遇三次以上弹窗干扰后对应用的信任评分会下降约40%65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/932783API
curl https://kongchang.com/api/v1/knowledge/claims/932783MCP
get_claim(id=932783)