AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

ProfBench: Multi-Domain Rubrics requiring Professional Knowledge to Answer and Judge

Zhilin Wang, Jaehun Jung, Ximing Lu, Shizhe Diao, Ellie Evans, Jiaqi Zeng, Pavlo Molchanov, Yejin Choi, Jan Kautz, Yi Dong

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Knowledge-augmented question answering · 22 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
provide grounding documents instead of LLM only (ablation)o3 · ProfBenchscore (%): 52.0 -> 61.4Improved
provide grounding documents instead of LLM only (ablation)o4-mini · ProfBenchscore (%): 46.3 -> 58.2Improved
add web search to prompt without documents (ablation)o3 · ProfBenchscore (%): 52.0 -> 56.1Improved
raise GPT-5 reasoning effort from minimal to high (ablation)GPT-5 · ProfBenchscore (%): 61.1 -> 65.9Improved
raise GPT-5 verbosity from low to medium (ablation)GPT-5 · ProfBenchscore (%): 62.2 -> 65.2Improved
raise GPT-5 verbosity from medium to high (ablation)GPT-5 · ProfBenchscore (%): 65.2 -> 66.3No clear effect
use thinking Qwen3-30B-A3B instead of instructQwen3-30B-A3B-2507 · ProfBenchscore (%): 49.3 -> 44.6Hurt
increase judge reasoning effort (ablation)GPT-5 as judge · ProfBench judgeoverall (MF1 - BI): minimal 71.9 -> high 76.0; med 76.7Mixed
use mixed reasoning effort by domain for gpt-oss-120b judge (ablation)gpt-oss-120b · ProfBench judgeoverall (MF1 - BI): 76.7 (low) -> 78.2Improved
scale model size within familyGPT-5 family · ProfBenchscore (%): GPT-5-nano 50.1 -> mini 60.3 -> GPT-5 65.9Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy