Research & Benchmarks - Page 5 of 34
Academic AI research, performance benchmarks, scientific breakthroughs, and peer-reviewed studies advancing artificial intelligence frontiers.
680 articles
View complete article list
Google's ME-POIs Adds "How a Place Is Used" to POI Embeddings
Two coffee shops can sit on the same block, carry the same category tag, and produce nearly identical text embeddings, while one turns over commuters in five minutes and the other holds customers for an hour and a half.
NVIDIA's Groq 3 LPX Enters Full Production for AI Agents
NVIDIA's Groq 3 LPX chip has moved into full production, and the company is pairing it with its Vera Rubin NVL72 rack-scale system to chase a specific problem: getting AI agents to generate tokens fast enough for real-world use.
Harvey Launches AI for M&A Diligence, Contract Review Over 10,000 Docs
Harvey announced a research preview on August 20, 2026, called Harvey Tenet, its first post-trained model built for legal work that spans thousands of documents at once, the kind of contract review or M&A diligence job that used to eat a junior...
A.I. Agent Adoption Grows More Slowly Than AI Itself
Gartner expects more than 40% of today's agentic AI projects to be scrapped before 2028.
AI Labs Lack Plans to Contain Rogue Models, Documents Show
Guidelight AI Standards graded five frontier AI labs on a question none of them like to answer directly: what happens the moment one of their models gets caught trying to slip out of human control. OpenAI came out on top of the ranking.
AI Agents Succeed or Fail Based on Their "Skills
Give an AI agent a "skill," a short cheat sheet of steps and warnings for a task, and it often does better. That much AI developers already suspected. What nobody had nailed down was why, or when the trick stops working.
Ignoring Human Beliefs Leads AI to Predict Wrong Actions, Study Finds
Sora can render a coffee cup vanishing into a cabinet with perfect physics. What it can't tell you is what the person who owns that cup will do next, because it has no idea they didn't see it move.
RadixAttention Speeds First Tokens by Reusing Cached KV States
A Llama-3 8B model holding a 100,000-token conversation in memory needs close to 12.8 GiB just for its KV cache, before a single additional request gets queued.
Deepseek Targets Visual Agents With New Experimental Flash Model
Deepseek pushed out V4-Flash-Vision-Exp this week, an experimental multimodal model that bolts image understanding onto its existing text-only V4-Flash system.
Meta Pays Microsoft Hundreds of Millions Annually for AI
Meta burns through trillions of tokens a week on Microsoft's Azure cloud, and according to a Bloomberg report, that habit costs the company hundreds of millions of dollars annually.
Anthropic's Claude Expands Into Protein Design
Four days ago, Dario Amodei posted on X that Anthropic's biology work was still months from producing its first "early glimmers." That timeline just moved up.
OpenAI Revokes Access to Limited Cyber Program Daybreak Blue
Wednesday brought a wave of complaints from security researchers who found themselves locked out of Trusted Access for Cyber, OpenAI's vetted program for using its top AI models with fewer cybersecurity restrictions than regular ChatGPT accounts...
GLM-5.3 Jumps 246 Points on Key Benchmark, Ranks Second to Claude Opus
Z.ai's GLM-5.3 just closed the gap with the best closed model on the market. On the GDPval-AA v2 benchmark, which measures agentic task performance, the model's Elo score climbed from 1,524 to 1,770.
OpenAI Pauses Astra Model Work Over Safety Concerns
OpenAI has put a hold on parts of its work on Astra, the company's model project, citing safety concerns that haven't been fully detailed publicly.
New LLM Agent Achieves 100% Accuracy on FDA-Based Clinical Trial Benchmark
Ask five leading language models to write a single clinical trial dataset from scratch, and none of them will get it right.
Cartesia's Sonic-3.6 TTS Model Adds Natural Pauses and Hinglish Code-Switching
Cartesia put out Sonic-3.6 this week, the latest version of its real-time text-to-speech model and a follow-up to Sonic-3.5 from three months back.
OpenAI slows model development amid rising cybersecurity risks
OpenAI put a two-week hold on reinforcement learning training this fall, and its largest planned frontier RL run is still sitting on ice.
New Benchmark Tests AI Search APIs on 900 Research Questions
Artificial Analysis has a new way to grade the search tools that AI agents lean on when they go looking for answers online.
New AI Art Study Finds Images Can't Be Traced to Training Data
Somewhere between an artist's original painting and an AI-generated image of it, the trail runs out.
Optima's New AI Benchmark Lets Users Test Models With Their Own Data
Artificial Analysis, the research group behind independent LLM evaluations and benchmarks like GDPval-AA and AA-Briefcase, has launched a new platform called Optima.
Featured Resources & Reviews
No Code MBA Course Review
Learn to build AI-powered apps without coding. Our comprehensive review of No Code MBA's course.
AI Tools & Resources
Curated collection of AI tools, courses, and frameworks to accelerate your AI journey.
Weekly AI Digest
Get the week's most important AI news delivered to your inbox every week.