i10X News
AI news that matters
Models, agents, deals and policy — analysed daily by the i10X newsroom.
Latest
-
A 3B Model Hits 94.3 on AIME26 and 80.2 on LiveCodeBench
-
Turn Off Thinking Mode and Models Still Reason 99.9% of the Time
-
Even Top Agents Finish Only ~30% of Real Startup Product Workflows
-
No Single Agent Memory Architecture Wins Everywhere, Systems Study Finds
-
Coding Agents Can’t Build the Agent You Actually Want — Yet
-
Coding Agents Reach the Right File but Still Miss the Lines That Matter
-
Agent “Skills” Mostly Anchor Procedures, Not Inject Knowledge — and Retrieval Breaks Them
-
LLMs Can Build Their Own Agent Harnesses — But Still Trail Human Ones on Code
-
Hybrid Computer-Use Agents Top Out at 41% When GUI and CLI Must Work Together
-
Scientific AI Agents Finish Only 21% of End-to-End Workflows — and Often Claim Success Anyway
Editor's picks
Try the models in the news
Compare side by side ›Anthropic & Claude
All Anthropic & Claude news ›OpenAI & ChatGPT
All OpenAI news ›Google & Gemini
All Google news ›Agents
All Agents news ›Models
All Models news ›New studies
All studies ›
Society & EconomyGoogle Study: Structured AI Agent Prepares Patients for Doctor Visits Better Than a Plain ChatbotPreprint · Nov 2
Society & EconomyAI Hiring Screeners Favor Cover Letters Written by Certain AI Models, Study FindsPreprint · Nov 2
Agent ResearchOne in 17 AI Agent Skills Tells Agents to Skip Human Oversight, Study FindsPreprint · Oct 30
Benchmarks & EvalsAI Security Agents vs. SAST: Snyk Benchmark Finds They Catch Different BugsPreprint · Oct 30