List your company

Enterprise Workflow Automation on WorkArena (test)

45.1M&D ScoreMem-π
31.37234.93638.542.064May 20, 2026

Evaluation results

MethodLinks
Mem-πBase Agent=gpt-5.4-mini, Memory Backbone=Qwen-2.5-7B-Instruct2026.0545.1—70.613.685.350.3
Mem-π (Stage 1)Base Agent=gpt-5.4-mini, Memory Backbone=Qwen-2.5-7B-Instruct, Training Stage=Stage 12026.0541.6—65.8981.546.6
MemRLBase Agent=gpt-5.4-mini2026.0536.5—63.610.680.546.1
Memory-R1Base Agent=gpt-5.4-mini, Memory Backbone=Qwen-2.5-7B-Instruct2026.0535.4—62.310.580.544.3
Mem0Base Agent=gpt-5.4-mini, k=12026.0533.9—61.89.579.144.1
RAGBase Agent=gpt-5.4-mini, k=12026.0533.6—60.6978.142.6
Base AgentBase Agent=gpt-5.4-mini2026.0531.9—55.711.676.942