List your company
ResearchFormerly WizWand

General Knowledge on MMLU (Accuracy)

91.2MMLU General Knowledge AccuracyQwen3.5-35B-A3B
74.019278.479682.9487.4004Dec 8, 2025Jan 12, 2026Feb 16, 2026Mar 23, 2026Apr 27, 2026Jun 1, 2026Jul 7, 2026
Updated 3mo ago

Evaluation results

MethodLinks
Qwen3.5-35B-A3B#Token=9002026.0491.2
DeepSeek V3.2Evaluation Mode=Chat2025.1291.1
GLM 4.6Evaluation Mode=Chat2025.1290.7
LongCat-Flash ChatEvaluation Mode=Chat2025.1289.7
LongCat-Flash Exp-ChatEvaluation Mode=Chat2025.1289.6
JoyAI-LLM Flash#Token=2002026.0489.1
Qwen3.5-35B-A3B-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0488.4
GLM-4.7-Flash-T#Token=20002026.0488.2
Qwen3-30B-A3B#Token=2002026.0488
Kimi-K2 Base# Shots=5-shot, # Activated Params=32B, # Total Params=1043B2026.0187.8
DeepSeek-V3.2 Exp Base# Shots=5-shot, # Activated Params=37B, # Total Params=671B2026.0187.8
DenserBackbone=Qwen3-32B-think2025.1287.8
Process SupervisionBackbone=Qwen3-32B-think2025.1287.7
DeepSeek-V3.1 Base# Shots=5-shot, # Activated Params=37B, # Total Params=671B2026.0187.4
Reflection-CoTBackbone=Qwen3-32B-think2025.1287.3
MiMo-V2-Flash Base# Shots=5-shot, # Activated Params=15B, # Total Params=309B2026.0186.7
Self-VerificationBackbone=Qwen3-32B-think2025.1286.4
Qwen3-Next-80B-A3B#Token=4002026.0486.4
DenserBackbone=Qwen3-32B2025.1286.2
Tree-of-ThoughtBackbone=Qwen3-32B-think2025.1286.1
N-3-Super 120B-A12B-BaseShots=52026.0486.01
Process SupervisionBackbone=Qwen3-32B2025.1285.9
Self-ConsistencyBackbone=Qwen3-32B-think2025.1285.6
Reflection-CoTBackbone=Qwen3-32B2025.1285.5
Qwen-3-30B-A3BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0285
Think-to-ThinkBackbone=Qwen3-32B-think2025.1284.9
Self-VerificationBackbone=Qwen3-32B2025.1284.7
JoyAI-LLM Flash-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0484.7
Llama-3.3-70BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0284.6
Tree-of-ThoughtBackbone=Qwen3-32B2025.1284.3
Qwen-3-32BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0284
Self-ConsistencyBackbone=Qwen3-32B2025.1283.8
Qwen3Parameters=8B, Architecture=AR2026.0483.5
Chain-of-ThoughtBackbone=Qwen3-32B-think2025.1283.2
Think-to-ThinkBackbone=Qwen3-32B2025.1283.1
NBDiffParameters=7B2026.0482.9
I-DLMParameters=8B2026.0482.4
Qwen3-30B-A3B-BaseEvaluation Framework=OpenCompass, Decoding=Greedy2026.0482.1
DenserBackbone=Qwen3-8B-think2025.1281.8
Qwen3-30B-A3BSparsity Level=Base, K=82026.0581.8
Process SupervisionBackbone=Qwen3-8B-think2025.1281.7
Chain-of-ThoughtBackbone=Qwen3-32B2025.1281.5
Reflection-CoTBackbone=Qwen3-8B-think2025.1281.3
Qwen-3-14BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0281.2
Ling-flash base-2.0Shots=52026.0481
GLM-4.5 Air-BaseShots=52026.0481
Gemma-3-27BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0280.4
DenserBackbone=Qwen3-14B-no-think2025.1280.3
DenserBackbone=Qwen3-8B2025.1280.2
Process SupervisionBackbone=Qwen3-14B-no-think2025.1280.2
OLMo-3.1-32BOpenness=Fully-open, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0280.1
Tree-of-ThoughtBackbone=Qwen3-8B-think2025.1280.1
Self-VerificationBackbone=Qwen3-8B-think2025.1280.1
BEAMSparsity Level=Mid Sparsity, beta=0.012026.0580.09
Process SupervisionBackbone=Qwen3-8B2025.1280
Reflection-CoTBackbone=Qwen3-14B-no-think2025.1279.8
Reflection-CoTBackbone=Qwen3-8B2025.1279.6
Self-ConsistencyBackbone=Qwen3-8B-think2025.1279.6
DenserBackbone=Qwen3-4B-think2025.1279.1
Think-to-ThinkBackbone=Qwen3-8B-think2025.1278.9
Self-VerificationBackbone=Qwen3-14B-no-think2025.1278.7
Process SupervisionBackbone=Qwen3-4B-think2025.1278.6
Self-VerificationBackbone=Qwen3-8B2025.1278.6
Tree-of-ThoughtBackbone=Qwen3-14B-no-think2025.1278.6
SDAR-8BScale=8B, Tokens=55B, FLOPs=2640, Decoding Strategy=greedy static decoding, Block Size=42026.0678.6
Tree-of-ThoughtBackbone=Qwen3-8B2025.1278.4
Top-K ReducedSparsity Level=Mid Sparsity, K=42026.0578.27
Reflection-CoTBackbone=Qwen3-4B-think2025.1278.2
Self-ConsistencyBackbone=Qwen3-14B-no-think2025.1278.1
Self-ConsistencyBackbone=Qwen3-8B2025.1277.9
MoE-DynamicSparsity Level=Mid Sparsity, phi=0.32026.0577.86
Self-VerificationBackbone=Qwen3-4B-think2025.1277.5
Chain-of-ThoughtBackbone=Qwen3-8B-think2025.1277.4
Think-to-ThinkBackbone=Qwen3-14B-no-think2025.1277.4
Mistral-3.2-24BOpenness=Open-weights, Regional Origin=European, Post-training=Instruction-tuned2026.0277.3
Think-to-ThinkBackbone=Qwen3-8B2025.1277.2
Tree-of-ThoughtBackbone=Qwen3-4B-think2025.1277
Qwen3-8BGen. Mode=AR, Avg TPF=1.002026.0776.93
DenserBackbone=Qwen3-4B2025.1276.9
Qwen3-8BParameters=8B2026.0176.8
Molmo2-8BParameters=8B2026.0176.6
TiDARParameters=8B2026.0476.6
Self-ConsistencyBackbone=Qwen3-4B-think2025.1276.5
Ministral3-8BGen. Mode=AR, Avg TPF=1.002026.0776.39
Process SupervisionBackbone=Qwen3-4B2025.1276.2
Gemma-3-12BOpenness=Open-weights, Regional Origin=Non-European, Post-training=Instruction-tuned2026.0276.1
BEAMSparsity Level=High Sparsity, beta=0.12026.0576.06
Chain-of-ThoughtBackbone=Qwen3-14B-no-think2025.1275.9
Reflection-CoTBackbone=Qwen3-4B2025.1275.8
Think-to-ThinkBackbone=Qwen3-4B-think2025.1275.8
Chain-of-ThoughtBackbone=Qwen3-8B2025.1275.8
Qwen3-4BParams=4B2025.1275.78
WeDLMParameters=8B2026.0475.5
BaseAllowed Domains=—2026.0575.4
Self-VerificationBackbone=Qwen3-4B2025.1275.2
SDAR-4BScale=4B, Tokens=55B, FLOPs=1320, Decoding Strategy=greedy static decoding, Block Size=42026.0674.9
PALETTEAllowed Domains=Illegal | Violence2026.0574.8
Tree-of-ThoughtBackbone=Qwen3-4B2025.1274.7
Nemotron-Labs-Diffusion-8BGen. Mode=AR, Avg TPF=1.002026.0774.68
Nemotron-Labs-Diffusion-8BGen. Mode=Diff., Avg TPF=2.062026.0774.68