List your company
ResearchFormerly WizWand

General Knowledge on MMLU-Pro (Accuracy)

86.7AccuracyQwen3.5-122B-A10B
8.294428.649749.00569.3603Sep 8, 2025Oct 28, 2025Dec 17, 2025Feb 5, 2026Mar 27, 2026May 16, 2026Jul 5, 2026
Updated 3mo ago

Evaluation results

MethodLinks
Qwen3.5-122B-A10B2026.0486.7
Qwen-3.5 27BSpeedup @32k=0.55x, Speedup @16k=0.5x2026.0486.1
SuperPrecision=BF162026.0783.8
Nemotron 3 Super2026.0483.73
SuperPrecision=NVFP42026.0783.5
Puzzle-75B-A9BPrecision=BF162026.0782.4
Puzzle-75B-A9BPrecision=NVFP42026.0782.2
Qwen3.5-35B-A3B#Token=40002026.0481.6
JoyAI-LLM Flash#Token=9002026.0481.6
GPT-OSS-120B2026.0481
Apriel-1.6Speedup @32k=1.0x, Speedup @16k=1.0x2026.0478.8
Qwen3-30B-A3B#Token=16002026.0478.7
Nemotron-3-Nano 30BSpeedup @32k=4.09x, Speedup @16k=2.8x2026.0478.3
Nemotron-Nano 12B v2Speedup @32k=5.85x, Speedup @16k=4.3x2026.0477.2
all-attentionSpeedup @32k=1.0x, Speedup @16k=1.0x2026.0476.8
Qwen3-Next-80B-A3B#Token=15002026.0476.7
Reg|Lklhd–18Speedup @32k=4.76x, Speedup @16k=2.2x2026.0476.3
Idealized|All–18Speedup @32k=1.99x, Speedup @16k=1.1x2026.0476.3
Reg|Lklhd–26Speedup @32k=2.85x, Speedup @16k=1.5x2026.0476.2
GLM-4.7-Flash-T#Token=62002026.0475.8
S1: Distil. Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.0474.5
Idealized|Lklhd–6Speedup @32k=6.2x, Speedup @16k=2.4x2026.0473.6
Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.0473.3
JoyAI-LLM Flash-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0473.1
QwenCheckpoint=SFTMaxOOD2025.0973
Falcon-H1R 7BSpeedup @32k=4.61x, Speedup @16k=3.4x2026.0472.5
Reg|Lklhd–13Speedup @32k=6.9x, Speedup @16k=2.7x2026.0469.3
Reg|Lklhd–10Speedup @32k=10.69x, Speedup @16k=4.2x2026.0468.2
Apriel-H1 15BSpeedup @32k=1.97x, Speedup @16k=1.9x2026.0467.5
QwenCheckpoint=SFTEnd2025.0966
QwenCheckpoint=RLEnd2025.0966
OLMo-Hybrid-Think 7BSpeedup @32k=2.51x, Speedup @16k=2.1x2026.0465.7
Qwen3-30B-A3B-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0461.7
Qwen3.5-35B-A3B-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0460.7
SDAR-8BScale=8B, Tokens=55B, FLOPs=2640, Decoding Strategy=greedy static decoding, Block Size=42026.0656.9
OPDLM-8BScale=8B, Tokens=0.066B, FLOPs=4.2, Decoding Strategy=greedy static decoding, Block Size=42026.0653.7
Qwen3Model Version=4B, Architecture=Dense, # Params=4.0B, # Tokens=36T2025.1051.4
SDAR-4BScale=4B, Tokens=55B, FLOPs=1320, Decoding Strategy=greedy static decoding, Block Size=42026.0650.9
OuroModel Version=1.4B R4, Architecture=LoopLM, # Params=1.4B, # Tokens=7.7T, recurrent steps=42025.1048.62
OPDLM-4BScale=4B, Tokens=0.076B, FLOPs=2.4, Decoding Strategy=greedy static decoding, Block Size=42026.0646.3
LLaMACheckpoint=SFTMaxOOD2025.0946
Dream-7BScale=8B, Tokens=580B, FLOPs=24360, Decoding Strategy=greedy static decoding, Block Size=42026.0643.3
Fast-dLLM-v2-7BScale=8B, Tokens=1B, FLOPs=42, Decoding Strategy=greedy static decoding, Block Size=42026.0641.5
MSA-CPTTraining Strategy=sparse continued pretraining2026.0639.1
MSA-PTTraining Strategy=from-scratch sparse pretraining2026.0638.8
FullTraining Strategy=Full-Attention baseline2026.0638.5
Qwen2.5Model Version=3B, Architecture=Dense, # Params=3.0B, # Tokens=18T2025.1037.87
Qwen3Model Version=1.7B, Architecture=Dense, # Params=1.7B, # Tokens=36T2025.1037.27
LLaMACheckpoint=RLEnd2025.0937
LLaDA-8BScale=8B, Tokens=1500B, FLOPs=72000, Decoding Strategy=greedy static decoding, Block Size=42026.0637
LLaMACheckpoint=SFTEnd2025.0936
Gemma3Model Version=4B, Architecture=Dense, # Params=4.0B, # Tokens=4T2025.1034.61
Llama3.2Model Version=3B, Architecture=Dense, # Params=3.0B, # Tokens=9T2025.1033.34
Qwen2.5Model Version=1.5B, Architecture=Dense, # Params=1.5B, # Tokens=18T2025.1029.11
EMM2026.0628.1
MERGEvolveProtocol=Cross-domain generalization2026.0627.17
Pack of LLMs2026.0627.07
LoraHubProtocol=Cross-domain generalization2026.0627.03
Pack of LLMsProtocol=Cross-domain generalization2026.0626.97
Model SwarmsProtocol=Cross-domain generalization2026.0626.92
MERGEvolve2026.0626.63
EMMProtocol=Cross-domain generalization2026.0626.54
Model Swarms2026.0626.47
LoraHub2026.0626.15
Expert Fusion2026.0625.3
Llama3.2Model Version=1.2B, Architecture=Dense, # Params=1.0B, # Tokens=9T2025.1011.8
Gemma3Model Version=1B, Architecture=Dense, # Params=1.0B, # Tokens=2T2025.1011.31