List your company

Tool-use Agent on HLE-500

26.1Text-Only PerformanceAutoAgent
1.8688.15914.4520.741Mar 10, 2026

Evaluation results

MethodLinks
AutoAgentBackbone=gemini-3-pro2026.0326.117.724.2
AutoAgentBackbone=gemini-3-pro-thinking2026.0323.517.722.2
Direct AnswerBackbone=gemini-3-pro-thinking2026.0320.415.619.4
Direct AnswerBackbone=gemini-3-pro2026.0320.414.219
DeepAgentBackbone=gemini-3-pro-thinking2026.0315.88.814.2
ReActBackbone=gemini-3-pro-thinking2026.0312.411.512.2
DeepAgentBackbone=gemini-3-pro2026.0312.14.410.4
ReActBackbone=gemini-3-pro2026.038.59.78.8
AutoAgentBackbone=gpt-4o2026.035.47.15.8
DeepAgentBackbone=gpt-4o2026.034.95.35
ReActBackbone=gpt-4o2026.033.45.33.8
Direct AnswerBackbone=gpt-4o2026.032.84.43.2