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

Aggregate Deep Research Performance on SQA, ResearchQA, and DRB v2

68.6Average Score

OpenAI Deep Research

Updated 1mo ago

Evaluation Results

Method	Links
OpenAI Deep Research 2026.06		68.6
DEEPRUBRIC (140-step, ≈750h) 2026.06		68.3
DR Tulu-8B (1900-step, ≈9700h) 2026.06		68.2
GPT-5 + Search 2026.06		67.9
DEEPRUBRIC (75-step, ≈402h) 2026.06		67.1
DR Tulu-8B (1000-step, ≈6000h) 2026.06		66.5
Ai2 ScholarQA 2026.06		66.3
Gemini 3 Pro + Search 2026.06		63.5
Perplexity Deep Research 2026.06		61.6
DEEPRUBRIC (SFT) 2026.06		60.8
DR Tulu-8B (SFT) 2026.06		59.9
WebThinker-32B-DPO (report) 2026.06		53.8
WebThinker (report) 2026.06		52
Tongyi DeepResearch-30B-A3B 2026.06		51.3
WebExplorer-8B 2026.06		48
QwQ-32B + RAG 2026.06		47.7
Qwen3-8B + RAG 2026.06		43.3
Qwen3-8B + Search 2026.06		40.6
WebThinker-32B-DPO 2026.06		34.9
Search-R1-7B 2026.06		19.9
ASearcher-Web-7B 2026.06		18