🏆 Top 10 MOE — Artificial Analysis

Fusion Small (4B-40B) + Medium (40B-150B) — tri global par Intelligence Index

Scrapé le mercredi 8 juillet 2026 à 18:30:21 UTC

10
Modèles MoE
Qwen3.5 122B A10B
Meilleur MoE
32.28
Score Top
#5
Créateurs

Classement unifié Top 10 Mixture of Experts

activeParams < passiveParams × 0.5
#ModèleCréateurIntelligenceParamètresContexte
1 Qwen3.5 122B A10B Alibaba 32.28 Active: 10.0B / Total: 115.0B 262K
2 Qwen3.6 35B A3B Alibaba 31.65 Active: 3.0B / Total: 33.0B 262K
3 Gemma 4 26B A4B Google 25.69 Active: 3.8B / Total: 21.4B 256K
4 NVIDIA Nemotron 3 Super NVIDIA 25.41 Active: 12.7B / Total: 107.9B 1.0M
5 gpt-oss-120b (high) Google 23.83 Active: 5.1B / Total: 111.9B 131K
6 Qwen3.5 35B A3B Alibaba 23.38 Active: 3.0B / Total: 33.0B 262K
7 Nemotron Cascade 2 30B A3B NVIDIA 21.25 Active: 3.0B / Total: 28.6B 1.0M
8 Qwen3 Coder Next Alibaba 21.11 Active: 3.0B / Total: 76.7B 256K
9 Mistral Small 4 Mistral AI 19.56 Active: 6.5B / Total: 112.5B 256K
10 HyperNova 60B 2605 – 17.8 Active: 4.8B / Total: 53.9B 131K