https://aitop.news/item/49302 AITOP en 2026-10-24T18:57:37.000Z Study Finds MLLM Judges Show Cultural Bias Under Ambiguity https://aitop.news/item/49298 AITOP en 2026-10-24T18:19:10.000Z Y Combinator Paper Club breaks down the full AI inference stack https://aitop.news/item/49297 AITOP en 2026-10-24T18:02:41.000Z vLLM Matches DeepInfer on DeepSeek-V4.1-Flash Decode https://aitop.news/item/49296 AITOP en 2026-10-24T17:50:16.000Z Researchers Make Skill Granularity a Runtime Decision for Coding Agents https://aitop.news/item/49274 AITOP en 2026-10-24T13:02:36.000Z EC²-Bench benchmarks edge-cloud collaboration for long-horizon agents https://aitop.news/item/49267 AITOP en 2026-10-24T11:43:14.000Z Billion-context: layered compression lets agents restore billion-token history https://aitop.news/item/49252 AITOP en 2026-10-24T10:20:19.000Z Micron Maps Memory and Storage Roadmap Through 2030 https://aitop.news/item/49242 AITOP en 2026-10-24T09:32:24.000Z University of Chicago Paper Finds LLMs Misreport Their Own Cruxes https://aitop.news/item/49233 AITOP en 2026-10-24T08:21:09.000Z Micron G11 NAND doubles density over G9 with 400-layer stacks https://aitop.news/item/49229 AITOP en 2026-10-24T08:02:15.000Z Moonshot Launches Kimi K3.1 Pro With 1M Context https://aitop.news/item/49223 AITOP en 2026-10-24T07:31:34.000Z OpenAI Codex Adds Mid-Conversation Reasoning Switching and CLI Updates https://aitop.news/item/49206 AITOP en 2026-10-24T05:32:06.000Z Cohere Launches Confidential Computing Version of Model Vault https://aitop.news/item/49171 AITOP en 2026-10-24T00:33:13.000Z Meta and Carnegie Mellon RL-trained 8B Qwen3 beats 402B Maverick on tool picking https://aitop.news/item/49132 AITOP en 2026-10-23T19:39:00.000Z CodeEvolve: Salesforce Research's LLM Evolutionary System for Code Enhancement https://aitop.news/item/49125 AITOP en 2026-10-23T18:53:38.000Z Cloudflare Origin Range Requests cut R2 p99 latency in half https://aitop.news/item/49123 AITOP en 2026-10-23T18:52:26.000Z Google Cloud AI Research Releases PLE Engineering Agent Benchmark https://aitop.news/item/49121 AITOP en 2026-10-23T18:47:54.000Z NVIDIA expands Cosmos tools for physical AI development https://aitop.news/item/49111 AITOP en 2026-10-23T18:25:26.000Z EdgeRock-4B open-weight model targets fixed-size memory reasoning https://aitop.news/item/49102 AITOP en 2026-10-23T17:54:15.000Z Amazon SageMaker AI guide to multi-turn reinforcement learning https://aitop.news/item/49078 AITOP en 2026-10-23T17:15:08.000Z Moonshot benchmark tweet sparks comparison with GLM 5.3 https://aitop.news/item/49076 AITOP en 2026-10-23T17:09:22.000Z Multi-Agent Systems Critique: Subagent Boundaries as Attention Masks https://aitop.news/item/49070 AITOP en 2026-10-23T17:02:18.000Z Tencent Hunyuan releases Hy-Nomas open source interactive video world model https://aitop.news/item/49066 AITOP en 2026-10-23T16:40:17.000Z AWS CDK construct automates SageMaker endpoint auto scaling updates https://aitop.news/item/49047 AITOP en 2026-10-23T15:33:29.000Z SenseTime ships RoboRSI-27B-Preview for embodied robotics https://aitop.news/item/49046 AITOP en 2026-10-23T15:32:13.000Z AWS open-sources TorchNeuron PyTorch backend for Trainium https://aitop.news/item/49033 AITOP en 2026-10-23T14:35:09.000Z Meta trains 8B Qwen3 model to beat larger LLMs at tool selection https://aitop.news/item/48968 AITOP en 2026-10-23T09:05:44.000Z TinyFish Details Agentic Search Architecture for Production AI Agents https://aitop.news/item/48941 AITOP en 2026-10-23T07:58:37.000Z China's domestic daily token volume hits 500 trillion in June https://aitop.news/item/48917 AITOP en 2026-10-23T06:04:05.000Z DeepSeek's Model Naming Scheme Points to Research Roadmap https://aitop.news/item/48916 AITOP en 2026-10-23T06:03:33.000Z COMPUTE Platform Shares 3 Design Lessons for AI Agent Infrastructure https://aitop.news/item/48914 AITOP en 2026-10-23T05:58:04.000Z Meta paper: RSD lets agents learn from unlabeled runs https://aitop.news/item/48882 AITOP en 2026-10-23T04:04:01.000Z Smaug-Nemotron-3.5 released as open-source model for single-machine agents https://aitop.news/item/48866 AITOP en 2026-10-23T03:25:20.000Z MiniMax M3.1 Flash Runs in Production on vLLM https://aitop.news/item/48863 AITOP en 2026-10-23T03:15:46.000Z Ant open-sources Ling-3.0-flash-Sante medical AI model https://aitop.news/item/48850 AITOP en 2026-10-23T02:54:26.000Z Study Finds Speculative Decoding Acceptance Does Not Decay Over Long Chain-of-Thought https://aitop.news/item/48849 AITOP en 2026-10-23T02:54:16.000Z Behavioral Counterfactuals Method Traces How Prompts Sway Model Reasoning https://aitop.news/item/48844 AITOP en 2026-10-23T02:53:28.000Z MiniCPM Spark Program S1 concludes with 16 winning projects https://aitop.news/item/48834 AITOP en 2026-10-23T02:49:09.000Z Simulated Control Evals Predict AI Safety With Lower Error https://aitop.news/item/48784 AITOP en 2026-10-23T01:46:03.000Z MiniMax M3.1 Flash lands on Together AI https://aitop.news/item/48783 AITOP en 2026-10-23T01:43:46.000Z SEA-HELM Audio Benchmark Targets Southeast Asian Speech Understanding Gaps https://aitop.news/item/48764 AITOP en 2026-10-23T01:03:04.000Z Sakana AI launches Aquarium agent platform for enterprise decisions https://aitop.news/item/48754 AITOP en 2026-10-23T00:21:06.000Z Artificial Analysis Coding Agent Index v2.0: Claude Opus 5.5 Leads https://aitop.news/item/48747 AITOP en 2026-10-23T00:03:15.000Z PyTorchCon 2026 Draws 3,500+ Attendees and Adds Nine Members https://aitop.news/item/48729 AITOP en 2026-10-22T22:56:53.000Z OpenAI makes Decisions API generally available https://aitop.news/item/48727 AITOP en 2026-10-22T22:51:50.000Z AA-PersonalityBench links model slop to intelligence gaps https://aitop.news/item/48708 AITOP en 2026-10-22T21:36:34.000Z Claude API Now Available Through Google Cloud Accounts https://aitop.news/item/48696 AITOP en 2026-10-22T20:34:22.000Z AWS Priming Converts Transformers Into Efficient Hybrid Models https://aitop.news/item/48691 AITOP en 2026-10-22T20:10:25.000Z Google DeepMind paper makes AI-text detector decisions interpretable https://aitop.news/item/48687 AITOP en 2026-10-22T20:03:42.000Z Anyscale Offers Hands-On Training on LLM Post-Training with SkyRL