List your company

Scenario-level policy synthesis on AppWorld normal (test)

98.2Task Goal Completion (TGC)HCL-GP
-3.92822.58649.175.614May 7, 2026

Evaluation results

MethodLinks
HCL-GPModel=Claude Sonnet 4.62026.0598.298.2
GPModel=Claude Sonnet 4.6, Note=our baseline2026.0596.496.4
Alibaba AgentRLModel=Qwen3-14B2026.0586.980.4
IBM CUGAModel=GPT-4.12026.0573.262.5
LOOPModel=Qwen2.5-32B2026.0572.653.6
Smolagents CodeModel=Claude Opus 4.52026.057060.4
OpenAI SoloModel=Claude Opus 4.52026.056856.6
Claude CodeModel=Claude Opus 4.52026.056656.6
ReAct ShortModel=Claude Opus 4.52026.056454.7
HCL-GPModel=GPT-OSS 120B2026.0562.562.5
ReActModel=Claude Opus 4.52026.056152.8
GPModel=GPT-OSS 120B, Note=our baseline2026.0500