Share your thoughts, 1 month free Claude Pro on usSee more

Deep Information Search and Synthesis on xbench DeepSearch

77.8Score

OpenAI-GPT-5-high

Updated 1mo ago

Evaluation Results

Method	Links
OpenAI-GPT-5-high 2026.02		77.8
Tongyi-DeepResearch 30B 2026.02		75
Tongyi DeepResearch 2026.03		75
OpenSeeker-v1-30B-SFT 2026.03		74
WebSailor-V2-30B-A3B 2026.02		73.7
WebSailor-V2-30B 2026.03		73.7
Minimax-M2 2026.02		72
WebLeaper-30B-A3B 2026.02		72
DeepSeek-V3.1-671B 2026.03		71.2
DeepSeek-V3.2 2026.02		71
GLM-4.6 2026.02		70
AgentCPM-Explore-4B 2026.02		70
Merged-Model-4B 2026.02		69.9
Kimi-Researcher 2026.02		69
Claude-4.5-Sonnet 2026.02		66
WebLeaper-30B 2026.03		66
WebSailor-V2-30B 2026.03		61.7
MiroThinker 8B 2026.02		60.6
WebSailor-32B 2026.03		53.3
DeepDive-32B 2026.03		51.8
ASearcher-QWQ-32B v2 2026.02		51.1
WebDancer-QwQ-32B 2026.02		38.3