List your company
ResearchFormerly WizWand

General Evaluation on LiveBench

46.83Accuracyphi-balancing
4.283615.329326.37537.4207May 14, 2026
Updated 5mo ago

Evaluation results

MethodLinks
phi-balancingModel=Moonlight-16B-A3B-Instruct2026.0546.83
ST-MoEModel=Moonlight-16B-A3B-Instruct2026.0542.21
Frozen checkpointModel=Moonlight-16B-A3B-Instruct2026.0519.57
phi-balancingModel=DeepSeek-V2-Lite2026.0519.42
phi-balancingModel=DeepSeek-V2-Lite2026.0519.34
ST-MoEModel=DeepSeek-V2-Lite2026.0518.8
phi-balancingModel=DeepSeek-MoE-Chat2026.0517.85
ST-MoEModel=DeepSeek-V2-Lite2026.0516.85
ST-MoEModel=DeepSeek-MoE-Chat2026.0514.62
phi-balancingModel=DeepSeek-MoE-Chat2026.0514.15
Frozen checkpointModel=DeepSeek-V2-Lite2026.0513.93
Frozen checkpointModel=DeepSeek-V2-Lite2026.0513.93
ST-MoEModel=DeepSeek-MoE-Chat2026.0513.79
Frozen checkpointModel=DeepSeek-MoE-Chat2026.055.92
Frozen checkpointModel=DeepSeek-MoE-Chat2026.055.92