List your company

Agentic Task Completion on AppWorld normal Hard (test)

39.7AccuracyNo compression
6.62815.21423.832.386Oct 1, 2025

Evaluation results

MethodLinks
No compressionAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=None2025.1039.711.959.11
ACON UTCOAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=Observation Compression2025.1031.89.147.5
ACON UTCOAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.1030.27.446.55
ACON UTAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.1028.67.656.37
PromptingAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=Observation Compression2025.10277.796.07
ACON UTAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=Observation Compression2025.10279.077.73
PromptingAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.1023.87.317.48
FIFOAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.1015.97.147.8
LLMLinguaAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.1015.98.098.59
LLMLinguaAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=Observation Compression2025.1014.310.299.92
RetrievalAgent=gpt-4.1, Compressor=gpt-4.1, Compression Type=History Compression2025.107.99.028.33