List your company

Creative Writing on WildBench

83.9WildBench ScoreBaseline
-2.52419.91342.3564.787Oct 15, 2025Nov 21, 2025Dec 29, 2025Feb 5, 2026Mar 14, 2026Apr 21, 2026May 29, 2026

Evaluation results

MethodLinks
BaselineModel=GLM-4.5-Air, Compression=Baseline2025.1083.9
REAPModel=GLM-4.5-Air, Compression=25%, Technique=Pruning2025.1083.1
EANModel=GLM-4.5-Air, Compression=25%, Technique=Pruning2025.1082.4
BaselineModel=Qwen3-30B-A3B, Compression=Baseline2025.1081.1
EANModel=Qwen3-30B-A3B, Compression=25%, Technique=Pruning2025.1081.1
FrequencyModel=Qwen3-30B-A3B, Compression=25%, Technique=Pruning2025.1080.7
M-SMoEModel=Qwen3-30B-A3B, Compression=25%, Technique=Merging2025.1080.5
REAPModel=Qwen3-30B-A3B, Compression=25%, Technique=Pruning2025.1080.4
FrequencyModel=GLM-4.5-Air, Compression=25%, Technique=Pruning2025.1079.3
HC-SMoEModel=GLM-4.5-Air, Compression=25%, Technique=Merging2025.1078.8
M-SMoEModel=GLM-4.5-Air, Compression=25%, Technique=Merging2025.1078.1
REAPModel=GLM-4.5-Air, Compression=50%, Technique=Pruning2025.1075.4
M-SMoEModel=Qwen3-30B-A3B, Compression=50%, Technique=Merging2025.1072.5
REAPModel=Qwen3-30B-A3B, Compression=50%, Technique=Pruning2025.1071.8
EANModel=Qwen3-30B-A3B, Compression=50%, Technique=Pruning2025.1070.2
EANModel=GLM-4.5-Air, Compression=50%, Technique=Pruning2025.1070.2
FrequencyModel=Qwen3-30B-A3B, Compression=50%, Technique=Pruning2025.1067.7
FrequencyModel=GLM-4.5-Air, Compression=50%, Technique=Pruning2025.1060.4
HC-SMoEModel=GLM-4.5-Air, Compression=50%, Technique=Merging2025.1059.3
HC-SMoEModel=Qwen3-30B-A3B, Compression=25%, Technique=Merging2025.1049.7
FullModel=ERNIE-21B, Pruning Ratio=0%, Protocol=Zero-shot2026.0347.9
SCOPEBase Model=Qwen3-8B, Task Type=Open-ended, # Seed Prompts=0, Rubric Source=Self-generated, Judge=Self-judge, Training Recipe=GRPO (60 steps), External Model Dependency=None2026.0546.1
FullModel=OLMoE-7B, Pruning Ratio=0%, Protocol=Zero-shot2026.0344.4
WebExplorerBase Model=Qwen3-8B, Task Type=Short-form QA, # Seed Prompts=~12K, Rubric Source=—, Judge=—, Training Recipe=SFT → GRPO (∼380 steps), External Model Dependency=DeepSeek-V3, Gemini 2.5 Flash2026.0541.7
REAPModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0340.5
AIMERModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0340.2
EANModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0339.6
M-SMoEModel=GLM-4.5-Air, Compression=50%, Technique=Merging2025.1039.1
SEERModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0338.1
RandomModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0337.7
WebThinker-R1Base Model=QwQ-32B, Task Type=Short-form QA, # Seed Prompts=~3K, Rubric Source=—, Judge=—, Training Recipe=DPO (2 iters), External Model Dependency=Qwen2.5-72B-Instruct2026.0536.5
DR TuluBase Model=Qwen3-8B, Task Type=Deep research, # Seed Prompts=~9K, Rubric Source=GPT-4.1, Judge=GPT-4.1-mini, Training Recipe=SFT → GRPO (1.9K steps), External Model Dependency=GPT-4.1, GPT-4.1-mini, GPT-52026.0536
FrequencyModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0335.2
RandomModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0329.7
REAPModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0326
EANModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0325.3
AIMERModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0323.6
MagnitudeModel=ERNIE-21B, Pruning Ratio=25%, Protocol=Zero-shot2026.0322.8
AIMERModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0321.7
EANModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0320.5
REAPModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0319.9
SEERModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0317.4
FrequencyModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0316.4
RandomModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.0315.8
SEERModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0314.1
FrequencyModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.0312.7
MagnitudeModel=ERNIE-21B, Pruning Ratio=50%, Protocol=Zero-shot2026.037.7
MagnitudeModel=OLMoE-7B, Pruning Ratio=25%, Protocol=Zero-shot2026.032
HC-SMoEModel=Qwen3-30B-A3B, Compression=50%, Technique=Merging2025.100.8