Research radar

arXiv preprints, filtered for developer impact.

Concise summaries of new AI papers, ranked for how likely they are to change how production systems are built. arXiv collection remains rate-limited.

Papers tracked

3,367

matching records

Shown

current page

Top radar

http://arxiv.org/abs/2610.06829v1

Code links

on this page

Paper	Authors	arXiv ID	Categories	Published	Code	Radar	Summary
CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling	Yifan Zhang, Yutong Dai, Viraj Prabhu	http://arxiv.org/abs/2610.06829v1	cs.CL, cs.AI	Oct 5, 2026	None	Score86	Open-source web agents are now strong enough to execute realistic browser tasks, but training...
PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data	Yaohui Zhang, Binxu Li, Haoyi Duan	http://arxiv.org/abs/2610.06825v1	cs.CL, cs.CV	Oct 5, 2026	None	Score83	Scientific figures often encode quantitative results that are not readily available in machin...
TasteVal: Measuring the Experimental Research Taste of AI Systems Against Human Experts	Oliver Jaffe, Dane Sherburn	http://arxiv.org/abs/2610.06824v1	cs.AI	Oct 5, 2026	None	Score86	We introduce TasteVal, a benchmark to evaluate the experimental research taste of frontier mo...
Make It Flow: Animating Water in 3D Gaussian Splatting Scenes	Kjell Keune, Florian Hahlbohm, Jannis Möller	http://arxiv.org/abs/2610.13860v1	cs.CV, cs.GR	Oct 9, 2026	None	Score78	Explicit 3D scene representations such as 3D Gaussian Splatting (3DGS) have recently enabled...
OmniCapBench: A Deep-Structured Evaluation Framework for Fine-Grained Audio-Visual Captioning	Zhongyu Yang, Jiale Tao, Ruitao Chen	http://arxiv.org/abs/2610.12458v1	cs.CV	Oct 8, 2026	None	Score78	Multimodal large language models (MLLMs) are rapidly evolving toward continuous audio--visual...
BrickBench: Evaluating Agentic Brick Design	Peter Kulits, Yiqing Xu, R. Kenny Jones	http://arxiv.org/abs/2610.12452v1	cs.AI, cs.CV	Oct 8, 2026	None	Score79	We propose BrickBench, a benchmark for agentic text-conditioned LEGO-set design. Given a prom...
One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts	Adrian Bulat, Yassine Ouali, Georgios Tzimiropoulos	http://arxiv.org/abs/2610.12448v1	cs.CV, cs.LG	Oct 8, 2026	None	Score77	In this work, we show that a single Transformer block, applied recurrently, can match the acc...
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception	Oskar J. Hollinsworth, Alex F. Spies, Tigist Diriba	http://arxiv.org/abs/2610.12445v1	cs.LG, cs.AI	Oct 8, 2026	None	Score81	Recent incidents have highlighted the challenge of monitoring LLM agents and the danger of mo...
SciExam for ENSO: Can AI Agents Build Climate Models?	Yinling Zhang, Langchen Liu, Dongbin Xiu	http://arxiv.org/abs/2610.10513v1	cs.AI, cs.LG	Oct 7, 2026	None	Score78	Language-model agents are increasingly asked to carry out open-ended scientific research, yet...
IdeaAnchor: Teaching LLMs to Turn Literature into Research Ideas	Ziyu Chen, Yilun Zhao, Jiashuo Sun	http://arxiv.org/abs/2610.08781v1	cs.CL, cs.AI	Oct 6, 2026	None	Score80	Scientific research often begins by synthesizing ideas from a set of related papers to identi...