📂 Category
Research & Benchmarks News Archive - Page 2 of 10
932 articles in this category • Page 2 of 10
- 101. Stanford, UC Berkeley TokenGate Speeds Long-Context Prefill Up to 11.3x
- 102. Anthropic co-founder Jack Clark moves self-improving AI prediction to 2027
- 103. Microsoft's Agensh Multi-Agent Harness Scales to 1,024 AI Agents
- 104. Kimi K3.1 Pro Rivals Claude Opus in Coding and Reasoning
- 105. Study: Fancy coding harnesses cost more but not better across 447 tasks
- 106. Kyutai Trains 1.2B-Parameter Model on 76,000 Hours of Video
- 107. OpenAI to stop training AI models on live internet over access risks
- 108. MIT's Martha Minow: AI threatens independent thinking in democracy
- 109. TinyFish Moves Planning Into Orchestrator's Unified Reasoning Loop
- 110. Agent4Sci testbed measures agents on 20 scientific ML benchmarks in 11 domains
- 111. Shenfeld et al. compare SFT and SDFT on three tasks for continual learning
- 112. Does Agentic RAG Retrieve Better Evidence? Test on 298 Questions
- 113. Researchers say AI agents learn skills by exploring without a goal
- 114. 8 NVIDIA DGX Spark Nodes Scale Frontier-Class AI Models With ConnectX-7 RoCE
- 115. 96% of AI teams anticipate needing faster archive retrieval for inference workloads
- 116. Perplexity, NVIDIA Researchers Introduce JOLT Joint Teacher-Student RL Method
- 117. NVIDIA Cosmos Data Factory Blueprints Speed Physical World Model Training
- 118. Shopify Gains 4.5x Speed Boost With NVIDIA AI Software
- 119. Jensen Huang Says Europe Can Lead Humanoid Robotics and Physical AI
- 120. Elevenlabs releases two speech-to-speech models for real-time AI conversations
- 121. Exa 3.0 Outperforms Rivals on 8 Benchmarks, Cuts Context Tokens by 95%
- 122. Barcelona Supercomputing Center Extends MareNostrum 5 for Spain-Based Startups
- 123. NVIDIA Separates Robot Safety Logic From Main Functions on the Same Computer
- 124. NVIDIA CUDA Tile Primitives Accelerate Mamba-2 With Faster exp2 Operations
- 125. Hugging Face policy chief says embedded AI evaluators won’t be easy solution
- 126. Hacker breached John Deere, SharkNinja with AI agent army across 100+ sites
- 127. Claude the AI, marooned on a virtual island, began to "play" and learn
- 128. New Benchmark Shows AI Agents Fail Long-Term Binary Choice Tests
- 129. AI models hide mistakes in work reports, but one prompt sentence can help
- 130. India reports rogue AI agents using leftover access, CERT-In says no hacking
- 131. China tightens AI export controls, bans malware-writing tools
- 132. Google Gemini 4 Argon scores 91.3% on ARC-AGI-2, nearing AI frontier
- 133. Cammarata's AI interpretability essay divides researchers over scale
- 134. Researchers Introduce AlphaDiana, a Modular System for Measuring AI Agents
- 135. Study Questions RLVR Paper Gains: Are Models Just Learning Answer Format?
- 136. Meta Muse and Apple HomeView surface US AI support for future viewpoints
- 137. Anthropic Gives METR Researchers Employee-Level Access to Systems
- 138. GPT-6-Astra solves 49 of 50 physics problems at USD 2.40 each
- 139. SpaceXAI Grok Voice 3.0 Ranks #1 on Artificial Analysis Speech-to-Speech Index
- 140. 17 Organizations Use AI to Accelerate Scientific Breakthroughs
- 141. Google: self-improving AI far off as fully delegated code changes rose to 11.9%
- 142. Researchers tracked rogue Chinese AI agents to a mailbox via hidden text
- 143. AI solves hundreds of math problems, triggering calls for limits
- 144. Anthropic Self-Replicating AI Warning: Gemma 4 Split Across 15 Devices
- 145. Floqer's AI agents research sales prospects so teams don't have to
- 146. OpenAI, Google AI models easily tricked into making nudes, researchers say
- 147. Harvard: men more optimistic and confident about AI than women
- 148. Gemini 4 Argon review: GraphWalks F1, DeepSWE, Vibe Code Bench
- 149. 406 scientists demand AI models cut off from internet over unpredictability
- 150. OpenAI employee arrested for allegedly planting bug in conference room
- 151. FBI arrests OpenAI employee accused of planting hidden recording device
- 152. Anthropic study finds small AI models can self-replicate on new devices
- 153. Z.ai GLM-Legal Scores 19.17% on Harvey LAB; 12.5-Point Lift Reproduced
- 154. OpenAI Launches Decisions API, Lets Models Plan Their Own Steps
- 155. Meta Launches Its First 'Frontier' AI Model, Muse Wonder
- 156. Google DeepMind's AlphaProof Nexus Solved 44 Conjectures, an Open Geometry Question
- 157. TII's Falcon-ASR Arabic Model Shows 20.92% WER, Trails Qwen3-ASR
- 158. AI agent teams waste tokens for barely measurable quality gains, research finds
- 159. Sakana AI's agentic LLM peer review catches 73% of core-claim errors
- 160. Anthropic severs internet connection for AI model safety tests
- 161. Anthropic AI Model Falsely Reported Homicide to Philadelphia Police
- 162. EmbeddingGemma 2 Unifies Text, Code, Images, Video and Audio
- 163. OpenAI Fires Three Safety Researchers for Policy Violations
- 164. NVIDIA's Astra AI Generates Code for SimReady Robot Assets
- 165. OpenAI disputes firings were over safety concerns after misconduct probe
- 166. AI Proofs Flood Math, Prompting Calls for OpenAI Boycott
- 167. AI Training Shows Hint of Future Robot Capabilities
- 168. Open TTS Leaderboard Adds Speaker Similarity Column, Ranks Top Multilingual Models
- 169. AWS AgentCore's default permissions bypass its own credential security advice
- 170. Single Attacker Breached South Korean Banks Using AI Tool ARTEX
- 171. Biohub Leads USD 1.8 Billion AI Push to Predict Cell Behavior
- 172. Google’s SynthID website now IDs AI-generated media
- 173. Weight-Loss Drugs May Slow Aging, Carbon Dioxide Batteries Emerge
- 174. OpenAI Publishes Solutions to 'Hundreds' of Open Math Questions
- 175. Researcher Counts Five to 10 New AI Chip Startups Every Year
- 176. Researchers Build Universal AI That Spans Physics to Biology
- 177. JEPA-Anything Applies Single Model Architecture to 7 Different Domains
- 178. Reflection AI's Beam Model Uses 501B Open Parameters for Coding Tasks
- 179. OpenAI consults mathematicians after AI missteps in math research
- 180. Researchers Track Chinese AI 'Agent Fleet' Performing Parallel Tasks
- 181. Yandex Sona Recommender Uses Dual Attention for Past and Recent Events
- 182. Ex-OpenAI Safety Researcher Criticizes Company's "Extreme Confidence
- 183. DeepMind Researchers Propose "Artificial Symbiotic Intelligence" as Singularity Alternative
- 184. Trillium Labs Launches With Focus on AI Fine-Tuning and RSI
- 185. NVIDIA's New DGX Spark 64GB Runs 100B-Parameter Models On-Device
- 186. OpenAI Fires Three, Fourth Departs Amid Safety Team Investigation
- 187. Business AI Usage Climbs 50% Since July Peak, Ramp Index Shows
- 188. AI solves accounting tasks flawlessly in benchmark with 160 tasks
- 189. Microsoft's New AI Transcribes 60 Languages in 100 Milliseconds
- 190. OpenAI Fires 3 Researchers Over Mishandled Information
- 191. OpenAI Links Stolen Model Attempts to China's Moonshot AI
- 192. AI Agents Tackle Thousands of Math Problems at Once
- 193. OpenAI Vows 'Responsible Disclosure' of Hack Details
- 194. AI Researchers Warn Superintelligence Is ‘Exactly as Dangerous as It Sounds’
- 195. NVIDIA's AI Development Constraint: Isolation as a Scaling Unit
- 196. BBC's Re-Released 2000 Exhibition Shows How We Imagined Internet's Future
- 197. Timnit Gebru Says AI's Real Issues Are Labor Exploitation and Data Documentation
- 198. Google Research Releases RRSI: Self-Improving AI Agents With Cost, Pruning Rules
- 199. AI Agents Assisted With Lab Grunt Work, Not Discovery
- 200. OpenAI Faces Scrutiny Over Hiring Ethics Amid Mathematician Criticism