List your company
ResearchFormerly WizWand

Reasoning on LiveBench

31.1AccuracyQwen2.5-7B
1.8769.46317.0524.637Oct 13, 2025Nov 21, 2025Dec 31, 2025Feb 9, 2026Mar 20, 2026Apr 29, 2026Jun 8, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen2.5-7BModel Backbone=Qwen2.5-7B2026.0631.1
PerSynStudent Model=Qwen2.5-3B2025.1022.3
CARStudent Model=Qwen2.5-3B2025.1020.9
Family-StrongStudent Model=Qwen2.5-3B2025.1020.4
MixStudent Model=Qwen2.5-3B2025.1019.3
StrongStudent Model=Qwen2.5-3B2025.1019.1
PerSynStudent Model=Qwen2.5-1.5B2025.1014.8
Family-StrongStudent Model=Qwen2.5-1.5B2025.1013.6
MixStudent Model=Qwen2.5-1.5B2025.1013.3
CARStudent Model=Qwen2.5-1.5B2025.1013.3
StrongStudent Model=Qwen2.5-1.5B2025.1012.8
CARStudent Model=Gemma-2-2B2025.1012.8
PerSynStudent Model=Llama-3.2-3B2025.1012.6
PerSynStudent Model=Gemma-2-2B2025.1012.4
Qwen2.5-0.5BModel Backbone=Qwen2.5-0.5B2026.0612.1
MixStudent Model=Llama-3.2-3B2025.1012
CARStudent Model=Llama-3.2-3B2025.1011.8
Family-StrongStudent Model=Gemma-2-2B2025.1011.6
Family-StrongStudent Model=Llama-3.2-3B2025.1011.1
MixStudent Model=Gemma-2-2B2025.1010.9
Uni-EnergyModel Backbone=LLaDa-8B, Decoding Strategy=Unified energy-based decoding2026.0610.9
StrongStudent Model=Llama-3.2-3B2025.1010.8
StrongStudent Model=Gemma-2-2B2025.1010.3
Uni-EnergyModel Backbone=Dream-7B, Decoding Strategy=Unified energy-based decoding2026.069.9
PerSynStudent Model=Qwen2.5-0.5B2025.109.8
Ind-EnergyModel Backbone=Dream-7B, Decoding Strategy=Independent energy-based decoding2026.069.3
CARStudent Model=Qwen2.5-0.5B2025.109
Inv-EnergyModel Backbone=Dream-7B, Decoding Strategy=Invariant energy-based decoding2026.068.8
Family-StrongStudent Model=Qwen2.5-0.5B2025.108.6
Ind-EnergyModel Backbone=LLaDa-8B, Decoding Strategy=Independent energy-based decoding2026.068.6
StrongStudent Model=Qwen2.5-0.5B2025.108.4
APDModel Backbone=Dream-7B, Decoding Strategy=Speculative decoding2026.068.2
MixStudent Model=Qwen2.5-0.5B2025.108.1
Dream-7BModel Backbone=Dream-7B, Decoding Strategy=Base2026.067.1
COREModel Backbone=Dream-7B, Decoding Strategy=Remask decoding2026.066.5
APDModel Backbone=LLaDa-8B, Decoding Strategy=Speculative decoding2026.066.3
Inv-EnergyModel Backbone=LLaDa-8B, Decoding Strategy=Invariant energy-based decoding2026.065.7
DAWNModel Backbone=LLaDa-8B, Decoding Strategy=Dependency decoding2026.064.8
LLaDa-8BModel Backbone=LLaDa-8B, Decoding Strategy=Base2026.064.1
COREModel Backbone=LLaDa-8B, Decoding Strategy=Remask decoding2026.063