List your company

Multimodal Maze Solving on MAZE

84Pass@1 AccuracyCoM
23.1638.95554.7570.545Feb 10, 2026

Evaluation results

MethodLinks
CoMBase Model=Gemini-2.0-Flash2026.0284
Direct I/OBase Model=Gemini-2.0-Flash2026.0276.5
MRPBase Model=Gemini-2.0-Flash2026.0276.5
Zero-shot CoTBase Model=Gemini-2.0-Flash2026.0276
ReActBase Model=Gemini-2.0-Flash2026.0271.5
Tree of ThoughtsBase Model=Gemini-2.0-Flash2026.0269.5
Meta-ReasonerBase Model=Gemini-2.0-Flash2026.0230.5
Chain of CodeBase Model=Gemini-2.0-Flash2026.0225.5