List your company

Reasoning on AIME 32K generation 25

74AccuracyOSCAR
51.463257.314163.16569.0159May 18, 2026

Evaluation results

MethodLinks
OSCARBackbone=Qwen3-32B, Generation Length=32K, KV-cache Mode=INT22026.05742.38
Original BF16Backbone=Qwen3-32B, Generation Length=32K, KV-cache Mode=BF162026.0572.5916
KittyBackbone=Qwen3-32B, Generation Length=32K, KV-cache Mode=INT22026.0569.262.39
OSCARBackbone=Qwen3-8B, Generation Length=32K, KV-cache Mode=INT22026.0566.672.38
Original BF16Backbone=Qwen3-8B, Generation Length=32K, KV-cache Mode=BF162026.056616
KittyBackbone=Qwen3-8B, Generation Length=32K, KV-cache Mode=INT22026.0559.672.39
KIVI-KV2*Backbone=Qwen3-32B, Generation Length=32K, KV-cache Mode=INT2 (Modified)2026.0559.052.26
KIVI-KV2*Backbone=Qwen3-8B, Generation Length=32K, KV-cache Mode=INT2 (Modified)2026.0557.672.26
KIVI-KV2Backbone=Qwen3-32B, Generation Length=32K, KV-cache Mode=INT22026.0557.412.25
KIVI-KV2Backbone=Qwen3-8B, Generation Length=32K, KV-cache Mode=INT22026.0552.332.25