List your company
ResearchFormerly WizWand

Accuracy on GPQA Diamond (Knowledge)

81.3Accuracy (GPQA Knowledge)Qwen3.5-9B
11.129.32547.5565.775Mar 10, 2026Mar 29, 2026Apr 18, 2026May 8, 2026May 27, 2026Jun 16, 2026Jul 6, 2026
Updated 3mo ago

Evaluation results

MethodLinks
Qwen3.5-9BParameters=9B, Variant=Thinking2026.0581.3
Kimi-K2.5Mode=Instant2026.0680.52
DeepSeek-V3.2Thinking Mode=nothink2026.0677.11
GPT-5.4Reasoning Mode=non-reasoning2026.0676.89
Qwen3.5-4BParameters=4B, Variant=Thinking2026.0576.8
Qwen3.5-Omni Flash 35B-A3BContext Length=256K2026.0776.4
Qwen3.5-4BActive / Total Parameters=4.0B / 4.0B, Sampling Settings=Recommended sampling settings2026.0576.2
Ling-2.6-1T2026.0676.17
Audex 30B-A3BContext Length=1M2026.0774.9
Qwen3-Omni 30B-A3B ThinkingContext Length=64K2026.0773.1
ZAYA1-8BActive / Total Parameters=0.7B / 8.0B, Sampling Settings=T=1.0, top-p=0.95, top-k disabled2026.0571
GLM-5Thinking Mode=non-thinking2026.0670.2
Qwen3-4B-Thinking-2507Active / Total Parameters=4.0B / 4.0B, Sampling Settings=Recommended sampling settings2026.0566.1
Step-Audio R1.1 33BContext Length=64K2026.0760.7
Qwen3-8BInference Mode=think2026.0360.1
SSA-LLM-8BInference Mode=think2026.0360.1
Mellum 2 (RL)Post-training Stage=RL, Parameters=2.5B/12B, Variant=Thinking2026.0557.6
Gemma-4-E4B-itActive / Total Parameters=4.0B / 8.0B, Sampling Settings=Recommended sampling settings2026.0557.4
Qwen3-14B + NGMModel Scale=14B, NGM Configuration=True, Decoding Settings=identical decoding settings2026.0552.02
Qwen3-8BInference Mode=no-think2026.0351.52
Qwen3-8B + NGMModel Scale=8B, NGM Configuration=True, Decoding Settings=identical decoding settings2026.0551.52
Qwen3-8BModel Scale=8B, NGM Configuration=False, Decoding Settings=identical decoding settings2026.0551.01
Qwen3-14BModel Scale=14B, NGM Configuration=False, Decoding Settings=identical decoding settings2026.0548.99
Audex 2BContext Length=128K2026.0747.2
Qwen3-4B + NGMModel Scale=4B, NGM Configuration=True, Decoding Settings=identical decoding settings2026.0546.46
Qwen2.5-72BParameters=72B2026.0546
Ministral-3-14BParameters=14B, Variant=Thinking2026.0546
SSA-LLM-8BInference Mode=no-think2026.0344.44
Qwen2.5-32BParameters=32B2026.0543.9
Qwen3-4BModel Scale=4B, NGM Configuration=False, Decoding Settings=identical decoding settings2026.0543.43
Qwen2.5-14BParameters=14B2026.0541.9
Mellum 2 (SFT)Post-training Stage=SFT, Parameters=2.5B/12B, Variant=Thinking2026.0539.9
FORGE-7B-NP-MATHParameters=7B, Configuration=NP-MATH2026.0538.9
Voxtral Small-24B 2507Context Length=128K2026.0737.9
FORGE-7B-NPParameters=7B, Configuration=NP2026.0537.4
InternLM3-8BParameters=8B2026.0536.9
FORGE-7B-MATHParameters=7B, Configuration=MATH2026.0533.8
Qwen2.5-7BParameters=7B2026.0533.3
Qwen3-1.7BModel Scale=1.7B, NGM Configuration=False, Decoding Settings=identical decoding settings2026.0532.83
Qwen3-1.7B + NGMModel Scale=1.7B, NGM Configuration=True, Decoding Settings=identical decoding settings2026.0531.31
Qwen2.5-3BParameters=3B2026.0530.3
OLMo-3-7BParameters=7B, Variant=Thinking2026.0529.3
Qwen3-0.6B + NGMModel Scale=0.6B, NGM Configuration=True, Decoding Settings=identical decoding settings2026.0529.29
Qwen3-0.6BModel Scale=0.6B, NGM Configuration=False, Decoding Settings=identical decoding settings2026.0526.77
LLama3.1-8bParameters=8B2026.0522.7
Original (T2T)Backbone=LLaDA2.1-mini, Inference Strategy=Text-to-Text2026.0421.72
T2MBackbone=LLaDA2.1-mini, Inference Strategy=Token-to-Mask, Remasking Strategy=LOWPROB, τ=0.3, Cmax=1, ρmax=0.252026.0421.72
Qwen2.5-7B-SFTParameters=7B, Configuration=SFT2026.0518.7
MiMo-Audio 7BContext Length=8K2026.0713.8