List your company
ResearchFormerly WizWand

Math on MATH 500 (Score)

96.6MATH 500 ScoreQwen3.5-9B (released)
19.22439.31259.479.488Jun 1, 2026Jun 3, 2026Jun 5, 2026Jun 7, 2026Jun 9, 2026Jun 11, 2026Jun 14, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen3.5-9B (released)Model Size=9B, Training Protocol=Released, Decoding Strategy=Native AR decoding2026.0696.6
Qwen3.5-4B (released)Model Size=4B, Training Protocol=Released, Decoding Strategy=Native AR decoding2026.0695.4
FLARE-9BModel Size=9B, Training Protocol=FLARE, Decoding Strategy=AR-Trust sampling2026.0695.2
Qwen3.5-9B + AR-SFTModel Size=9B, Training Protocol=AR-SFT, Decoding Strategy=Native AR decoding2026.0695
Qwen3.5-4B + AR-SFTModel Size=4B, Training Protocol=AR-SFT, Decoding Strategy=Native AR decoding2026.0694.4
FLARE-4BModel Size=4B, Training Protocol=FLARE, Decoding Strategy=AR-Trust sampling2026.0694.2
Qwen3.5-2B + AR-SFTModel Size=2B, Training Protocol=AR-SFT, Decoding Strategy=Native AR decoding2026.0685.8
FLARE-2BModel Size=2B, Training Protocol=FLARE, Decoding Strategy=AR-Trust sampling2026.0684.4
AMPOModel Category=Base and Expert Models2026.0677.4
TIESModel Category=Merged Models2026.0675.6
RAMModel Category=Merged Models2026.0675.4
DARE + TIESModel Category=Merged Models2026.0675.2
TAModel Category=Merged Models2026.0674.6
ISO-CModel Category=Merged Models2026.0674.6
TSV-MergeModel Category=Merged Models2026.0674.4
ISO-CTSModel Category=Merged Models2026.0674.2
RESMERGEModel Category=Merged Models2026.0673
ToolRLModel Category=Base and Expert Models2026.0672.6
Qwen3.5-2B (released)Model Size=2B, Training Protocol=Released, Decoding Strategy=Native AR decoding2026.0672.2
Qwen2.5-7B-InstructModel Category=Base and Expert Models2026.0671.4
REAPModel=ERNIE, Pruning Ratio=50%2026.0668
MANModel=ERNIE, Pruning Ratio=50%2026.0667.6
MemoryAgentModel Category=Base and Expert Models2026.0655.8
FullModel=DeepSeek, Pruning Ratio=0%2026.0629.8
FullModel=OLMoE, Pruning Ratio=0%2026.0622.2