Skip to content
VECTOR WIREAI INTELLIGENCE
UTC

AI Research Radar

Monday, November 16: 4 frontier papers out of 270 read.

4 frontier35 cleared270 papers read

Frontier

Papers rated as moving the field

How we choose

WebArena-Pro

On WebArena-Pro, Codex with GPT 6.1 Sol and Claude Code with Claude Opus 5.5 both reach 62.3%, the best among evaluated agents.

A Heterogeneous, Multimodal, Reproducible Benchmark for Web Agents

AgentsBenchmarkarXivPDF

Soteria

Soteria achieves 1.7 times the average task success rate of the best baselines across tool-use benchmarks, with no violations of policy specifications.

Safety

This week in research

Nov 10–16: 278 cleared, 18 frontier

Also cleared

31 papers, by field

Security & privacy13
All 13 papers in Security & privacy
Agents6
Robotics & embodied AI3
AI & society3
Safety & alignment2
Inference & efficiency2
Vision & multimodal1
Reasoning & math1