Unverified50% confidenceBenchmarkExact time
Enabling reasoning mode halved the refusal rates of two models.
1
Sources
50%
Confidence
Long-term
Relevance
10/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedHaving a second model perform a structured review against a rules checklist in the role of a critic can significantly reduce the probability of both models simultaneously producing the same hallucination.75% similarUnverifiedFor the two models tested, enabling maximum reasoning capability reduced the capitulation rate on deep puzzles from 19.2% and 12.5% directly to 0%.73% similarUnverifiedIn an ablation experiment, after replacing the model's latent thinking segments with segments from unrelated questions, accuracy dropped by 45%–51%, indicating that latent thinking carries substantive reasoning.71% similarUnverifiedLow-confidence requests first fail on the small model and then fall back to the large model with the same context for a second generation, causing such requests to incur the cost of two generations and resulting in double payment.70% similarUnverifiedDixon-Coles模型在双泊松模型基础上引入了低比分修正项，弥补了0:0和1:0等低分比赛中泊松假设的偏差70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1084339API
curl https://kongchang.com/api/v1/knowledge/claims/1084339MCP
get_claim(id=1084339)