SOTA Deep Research benchmarks and papers with code

Benchmarks

Dataset Name	SOTA Method	Metric
BrowseComp-ZH (BC-zh) original (test)		Pass@158.1	45	4d ago
xbench	RE-TRAC-30B-A3B	Accuracy83	30	4d ago
DeepResearch Bench official 100-task-subset 1.0	OAgents-DR	RACE Overall0.5076	24	4d ago
DeepResearch Bench	DualGraph	RACE Overall53.08	22	4d ago
BrowseComp	Kimi-K2.5-Agent	Score74.9	21	4d ago
BrowseComp-EN (BC-en) original (test)		Pass@149.7	20	4d ago
GAIA text-only original (test)	WebSailor-v2-30B-A3B (RL)	Pass@174.1	20	4d ago
BrowseComp+	Qwen3-235B (w/ Pensieve)	Accuracy55.33	19	4d ago
BrowseComp-zh		Accuracy66.6	18	4d ago
BrowseComp-zh		BrowseComp-zh Score81.3	16	4d ago
GAIA	THINKMERGE	Pass@151.46	16	4d ago
HLE		Accuracy51	16	4d ago
xbench-DS	DeepSeek-V3.1	Pass@171	15	4d ago
BrowseComp-ZH	OpenAI-o3	Pass@158.1	15	4d ago
BrowseComp	OpenAI-o3	Pass@150.9	15	4d ago
GAIA	OpenAI-o3	Pass@170.5	15	4d ago
XBench-DeepSearch original (test)	DeepSeek-V3.1	Pass@171	15	4d ago
GAIA	RE-TRAC-30B-A3B	Accuracy78.2	14	4d ago
WebWalkerQA original (test)	Tongyi-DeepResearch	Pass@172.2	14	4d ago
FRAMES	TOOLSELF	Accuracy56	14	4d ago
HLE text-only original (test)	Tongyi-DeepResearch	Pass@132.9	13	4d ago
DeepResearch Bench 1.0 (test)		Overall Score46.45	12	4d ago
ADR-Bench Finance & Law		Score Range2,535	9	4d ago
BC (test)	Our best agent	Mean Correct Answer Rate620	8	4d ago
GAIA (test)	Our best agent	Mean Correct Rate49.2	8	4d ago

Showing 25 of 40 rows