List your company
ResearchFormerly WizWand

Forecasting on Polymarket official (test)

64.99ScoreOurs (GPT-5.4-medium)
33.58241.73649.8958.044Jun 19, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Ours (GPT-5.4-medium)Agent Architecture=multi-agent (planner-solver-aggregator), Time Machine=on2026.0664.99
GPT-5.4-med-SC@5Agent Architecture=multi-agent, Self-Consistency=SC@5, Time Machine=on2026.0662.26
GPT-5.4-med-SC@3Agent Architecture=multi-agent, Self-Consistency=SC@3, Time Machine=on2026.0660.99
GLM-5 w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0659.24
GPT-5.4-medium w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0658.63
Gemini-3.1-Pro w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0658.45
Kimi-K2.5 w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0657.91
GPT-5.4-mediumMode=closed-book, Tool Usage=no tools2026.0653.76
Claude-4.6-Sonnet w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0648.21
DeepSeek-V4-Flash w/ toolsAgent Architecture=single-agent, Tool Usage=w/ tools, Time Machine=on2026.0648.12
Claude-4.6-SonnetMode=closed-book, Tool Usage=no tools2026.0647.97
Gemini-3.1-ProMode=closed-book, Tool Usage=no tools2026.0643.38
Kimi-K2.5Mode=closed-book, Tool Usage=no tools2026.0642.88
GLM-5Mode=closed-book, Tool Usage=no tools2026.0641.18
DeepSeek-V4-FlashMode=closed-book, Tool Usage=no tools2026.0634.79