Unverified50% confidenceBenchmarkExact time
The records made every model more aggressive—the total weight of betting, raising and all-in rose by 7 percentage points (luna), 12 (Haiku), 16 (Gemini) and 28 (Jev) respectively.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn the bet-sizing adjustment experiment, Jev's bet-size difference against opponents who fold to large bets was −54 percentage points, Gemini's was −33 percentage points, luna's was +16 percentage points, and Haiku's was +1 percentage point.66% similarUnverifiedTEMPEST outperforms the best classical model by 17.9 percentage points and the strongest triplet-loss baseline by 58.4 percentage points.61% similarUnverifiedOn the Deep Squeeze test, the old model scored 49%, while Gemini 3.7 Flash jumped to 65.3%.61% similarUnverifiedOn the Frontier Code programming benchmark, the old model scored 34.4%, while Gemini 3.7 Flash improved to 43.6%.60% similarUnverified在MoE模型中每个token只激活少量Expert（通常2-4个，Top-K路由），实际计算量可能仅相当于6-7B的Dense模型59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1047728API
curl https://kongchang.com/api/v1/knowledge/claims/1047728MCP
get_claim(id=1047728)