https://aitop.news/item/48506 AITOP en 2026-10-22T12:32:55.000Z JD ROAD MOPD refines multi-teacher distillation to cut gradient conflict https://aitop.news/item/48495 AITOP en 2026-10-22T11:33:47.000Z SemiAnalysis benchmarks Huawei Ascend 950PR against Nvidia H100 https://aitop.news/item/48472 AITOP en 2026-10-22T09:26:49.000Z vLLM Ships Day-0 Support for Sage-A1.5 https://aitop.news/item/48463 AITOP en 2026-10-22T09:16:00.000Z Tencent Hunyuan releases WorkForge for training work agents https://aitop.news/item/48461 AITOP en 2026-10-22T09:09:46.000Z Tencent Hunyuan releases WorkForge agent training framework https://aitop.news/item/48456 AITOP en 2026-10-22T08:36:07.000Z Muse Glimmer 30B: Meta's distilled agent model runs on one consumer GPU https://aitop.news/item/48402 AITOP en 2026-10-22T05:32:29.000Z Meta and Michigan State paper finds LLM-built reasoning graphs unreliable https://aitop.news/item/48391 AITOP en 2026-10-22T04:24:42.000Z DeepSeek Search ranks 17th on Artificial Analysis Search Index https://aitop.news/item/48375 AITOP en 2026-10-22T04:03:21.000Z Google Paper: OO-Harness Specialist Harnesses Outperform One Global Agent Harness https://aitop.news/item/48358 AITOP en 2026-10-22T03:03:05.000Z AI Discovers A-greement, a Faster Consensus Protocol https://aitop.news/item/48341 AITOP en 2026-10-22T02:49:52.000Z NVIDIA and Agile Robots launch humanoid reference platform Agile ONE SN https://aitop.news/item/48309 AITOP en 2026-10-22T02:41:35.000Z arXiv Paper Finds Formatting Cues Lost in Agent Document Ingestion https://aitop.news/item/48304 AITOP en 2026-10-22T02:40:54.000Z GitHub scales Git infrastructure on Azure for agents https://aitop.news/item/48292 AITOP en 2026-10-22T02:07:31.000Z Anthropic models lead completeness but trail OpenAI on medical benchmarks https://aitop.news/item/48239 AITOP en 2026-10-22T01:24:06.000Z Hermes Agent adds official Meshy plugin for 3D generation https://aitop.news/item/48221 AITOP en 2026-10-22T00:13:19.000Z Salesforce's Moirai 2.0 powers zero-shot forecasting in Data 360 https://aitop.news/item/48175 AITOP en 2026-10-21T21:05:47.000Z Claude API version 2026-10-21 released by Anthropic https://aitop.news/item/48158 AITOP en 2026-10-21T20:08:07.000Z AWS shows three ways to train Qwen3 on Trainium with torchtitan https://aitop.news/item/48157 AITOP en 2026-10-21T20:07:30.000Z PyTorchCon Demo Theater: 13 Sessions on Training, Inference and Agents https://aitop.news/item/48149 AITOP en 2026-10-21T19:43:49.000Z PyTorch Examines Linear Algebra Kernels Behind Faster Optimizers https://aitop.news/item/48133 AITOP en 2026-10-21T19:11:44.000Z Recursive Releases Flint, an Open-Weight 40B Coding Model https://aitop.news/item/48131 AITOP en 2026-10-21T18:54:42.000Z vLLM-TorchTPU brings native PyTorch serving to Google TPUs https://aitop.news/item/48130 AITOP en 2026-10-21T18:54:32.000Z Anthropic's Claude Security moves to continuous repository scanning https://aitop.news/item/48121 AITOP en 2026-10-21T18:26:12.000Z Kena's web index cuts agent search output tokens by 50% https://aitop.news/item/48108 AITOP en 2026-10-21T18:05:35.000Z AKV: Agents Manage Their Own KV Cache to Cut LLM Costs https://aitop.news/item/48104 AITOP en 2026-10-21T18:01:28.000Z FlyDSL kernel DSL now supports PyTorch on AMD MI350 GPUs https://aitop.news/item/48097 AITOP en 2026-10-21T17:37:57.000Z HeyGen Voice Turbo Tops Artificial Analysis Voice TTS Leaderboard https://aitop.news/item/48091 AITOP en 2026-10-21T17:25:24.000Z TorchTPU brings out-of-the-box PyTorch execution on TPUs https://aitop.news/item/48090 AITOP en 2026-10-21T17:23:26.000Z SGLang-TorchTPU Brings PyTorch Model Serving to Google TPUs https://aitop.news/item/48089 AITOP en 2026-10-21T17:13:22.000Z Crusoe Shows LLM Post-Training Platform at PyTorchCon 2026 https://aitop.news/item/48084 AITOP en 2026-10-21T17:07:33.000Z Cohere's Rachad Alao on open science for foundation models https://aitop.news/item/48080 AITOP en 2026-10-21T17:01:34.000Z Anyscale session covers Ray Serve LLM deployment and inference scaling https://aitop.news/item/48079 AITOP en 2026-10-21T16:55:26.000Z DeepSpeed adds AutoSP, Muon support and auto parallelism at PyTorchCon https://aitop.news/item/48076 AITOP en 2026-10-21T16:47:32.000Z vLLM adds Day-0 support for Recursive's Flint model https://aitop.news/item/48075 AITOP en 2026-10-21T16:43:24.000Z IBM Spyre joins PyTorch compiler and vLLM, Red Hat builds CI/CD https://aitop.news/item/48072 AITOP en 2026-10-21T16:31:23.000Z TorchTPU brings native PyTorch support for Google TPUs https://aitop.news/item/48071 AITOP en 2026-10-21T16:19:40.000Z PyTorchCon 2026 Keynote Maps Five Layers of the Agentic Stack https://aitop.news/item/48070 AITOP en 2026-10-21T16:19:34.000Z SF Research Launches EvoHarnessBench for Evolving Agent Harnesses https://aitop.news/item/48069 AITOP en 2026-10-21T16:13:35.000Z Scale AI Post-Trains NVIDIA Nemotron 3.5 Super VL https://aitop.news/item/48064 AITOP en 2026-10-21T16:05:32.000Z Databricks expands NVIDIA collaboration for open model fine-tuning https://aitop.news/item/48061 AITOP en 2026-10-21T15:55:38.000Z Claude Sonnet 5.5 and GPT-6 Astra tie on MLCR-AA leaderboard https://aitop.news/item/48041 AITOP en 2026-10-21T14:35:27.000Z Simular's Agent S4 evolves agent interfaces to boost OSWorld scores https://aitop.news/item/48039 AITOP en 2026-10-21T14:31:46.000Z QD-Rec Applies Open-Ended Evolution to Recommendation Tasks https://aitop.news/item/48033 AITOP en 2026-10-21T14:16:26.000Z Redis cuts AI application costs with LangCache, Agent Memory and Flex https://aitop.news/item/48012 AITOP en 2026-10-21T13:29:47.000Z Reasoning Models Show Generalization on Rare Text Encodings https://aitop.news/item/48009 AITOP en 2026-10-21T13:23:37.000Z Google Cloud to open Vera Rubin-based A5X instances next month https://aitop.news/item/48000 AITOP en 2026-10-21T13:07:53.000Z MiniCPM-V 4.7 runs 35B MoE on phones at 20.51 tokens/s https://aitop.news/item/47985 AITOP en 2026-10-21T12:29:38.000Z NVIDIA DSX POD Deploys AI Factories in 14 Days https://aitop.news/item/47956 AITOP en 2026-10-21T10:37:45.000Z dsh-context Tool Visualises Agent Context Usage for DSH https://aitop.news/item/47945 AITOP en 2026-10-21T10:05:30.000Z NVIDIA OpenShell Reference Architecture Sets Boundaries for AI Agents https://aitop.news/item/47916 AITOP en 2026-10-21T08:11:29.000Z Groq plans DRAM-stacked inference chip to fix SRAM capacity https://aitop.news/item/47887 AITOP en 2026-10-21T06:53:41.000Z Ant Group and vLLM Open-Source Humming Quantized Inference Kernel Library https://aitop.news/item/47847 AITOP en 2026-10-21T05:22:18.000Z Meta ProPerAgent Harness Lifts Proactive Agent Accuracy to 38% https://aitop.news/item/47839 AITOP en 2026-10-21T05:20:38.000Z Together Post-Training launches for custom open-model tuning https://aitop.news/item/47838 AITOP en 2026-10-21T05:20:36.000Z DeepInfer Release Draws Criticism Over Heterogeneous Hardware Support https://aitop.news/item/47832 AITOP en 2026-10-21T05:20:20.000Z Meta Publishes Technical Report on 30B Muse Glimmer Model https://aitop.news/item/47824 AITOP en 2026-10-21T05:09:25.000Z DeepSeek V4.1F shows 5x-7.4x decode edge on Ascend 950DT vs GB200 https://aitop.news/item/47786 AITOP en 2026-10-21T03:10:59.000Z Grab details layered defense for superapp crash stability https://aitop.news/item/47785 AITOP en 2026-10-21T03:05:10.000Z Tencent Cloud opens Yuanbao Search to developers at RMB 14 per 1,000 calls https://aitop.news/item/47784 AITOP en 2026-10-21T02:59:24.000Z Gemma-SEA-LION-Guard-v2 Adds Action Governance for Regional Agents https://aitop.news/item/47771 AITOP en 2026-10-21T02:25:28.000Z iFlytek Spark-X2.6-4B Launches With 1M-Token Context Under Apache 2.0 https://aitop.news/item/47770 AITOP en 2026-10-21T02:17:27.000Z Microsoft Paper Co-Evolves RL Harness and Qwen3.5 Model https://aitop.news/item/47757 AITOP en 2026-10-21T02:11:31.000Z Agent S4 Evolves Its Own Computer-Use Interfaces https://aitop.news/item/47756 AITOP en 2026-10-21T02:11:10.000Z R2S Fixes Myopic Teacher Scoring in Agentic On-Policy Distillation https://aitop.news/item/47723 AITOP en 2026-10-21T01:43:34.000Z DeepSeek Open-Sources DeepInfer Inference Framework for V4.1 Flash https://aitop.news/item/47718 AITOP en 2026-10-21T01:38:02.000Z SAR Metric Quantifies Vulnerability Monoculture Across Foundation Models https://aitop.news/item/47703 AITOP en 2026-10-21T00:49:39.000Z EAGLE-4 Hits 863.8 Tokens/s on 27B Models https://aitop.news/item/47696 AITOP en 2026-10-21T00:37:52.000Z FastH3-Omni Ships Realtime Video Generation with Open Weights https://aitop.news/item/47695 AITOP en 2026-10-21T00:37:43.000Z vLLM Matches DeepInfer Decode Performance on DeepSeek-V4.1-Flash https://aitop.news/item/47624 AITOP en 2026-10-20T19:28:17.000Z Levelsio shares prompt to build a private Alexa music skill https://aitop.news/item/47615 AITOP en 2026-10-20T19:05:10.000Z PyTorch-to-Spyre compiler pipeline explained by IBM team https://aitop.news/item/47609 AITOP en 2026-10-20T18:53:22.000Z Épi research agent sets 120 records on math problems https://aitop.news/item/47608 AITOP en 2026-10-20T18:49:12.000Z Google DeepMind paper finds self-improving agents forget old tasks https://aitop.news/item/47601 AITOP en 2026-10-20T18:30:55.000Z OpenAI adds 12-hour prompt caching to its API https://aitop.news/item/47600 AITOP en 2026-10-20T18:27:59.000Z NotebookLM Video Overview System Prompt Reveals Three-Stage Workflow https://aitop.news/item/47596 AITOP en 2026-10-20T18:19:09.000Z Mistral launches Voxtral 3 transcribe and speech models https://aitop.news/item/47588 AITOP en 2026-10-20T17:55:21.000Z Together cuts post-training price on B200 GPUs to $9 per hour https://aitop.news/item/47585 AITOP en 2026-10-20T17:52:05.000Z GPT-OSS-20B fine-tuned for multilingual reasoning on SageMaker HyperPod https://aitop.news/item/47579 AITOP en 2026-10-20T17:47:28.000Z Nvidia's Ujval Kapasi to Discuss Agentic PyTorch Performance on Vera Rubin https://aitop.news/item/47570 AITOP en 2026-10-20T17:41:19.000Z Databricks launches MemCurrent persistent memory for production agents https://aitop.news/item/47566 AITOP en 2026-10-20T17:26:46.000Z General Compute partners with Positron to offer Atlas inference systems https://aitop.news/item/47565 AITOP en 2026-10-20T17:23:16.000Z vLLM Project Shows KV Cache and Kernel Optimizations at PyTorchCon 2026 https://aitop.news/item/47561 AITOP en 2026-10-20T17:17:30.000Z Meta Details PyTorch-Native Stack at PyTorchCon 2026 https://aitop.news/item/47559 AITOP en 2026-10-20T17:17:09.000Z vLLM TorchTPU backend brings Google Cloud TPUs into vLLM ecosystem https://aitop.news/item/47556 AITOP en 2026-10-20T17:11:32.000Z Sara Hooker Keynote at PyTorchCon 2026 on Adaptive Intelligence https://aitop.news/item/47553 AITOP en 2026-10-20T17:07:16.000Z General Compute pairs GPUs with Positron chips for split inference https://aitop.news/item/47551 AITOP en 2026-10-20T17:00:56.000Z GPT-6 Astra lifts low-resource coding languages on MultiLangBench https://aitop.news/item/47550 AITOP en 2026-10-20T16:59:42.000Z TorchNeuron Brings Native PyTorch Backend to AWS Trainium https://aitop.news/item/47549 AITOP en 2026-10-20T16:59:31.000Z Ray evolves into distributed compute layer for RL frameworks https://aitop.news/item/47547 AITOP en 2026-10-20T16:56:35.000Z Zed benchmarks 25 approaches to MCP tool search https://aitop.news/item/47539 AITOP en 2026-10-20T16:44:17.000Z Shopify Details PyTorch and vLLM Continual Learning Loop at PyTorchCon 2026 https://aitop.news/item/47537 AITOP en 2026-10-20T16:36:54.000Z AWS Agent Registry Adds Runtime Tool Discovery for Agents https://aitop.news/item/47535 AITOP en 2026-10-20T16:35:50.000Z PyTorch keynote details torch.accelerator, FlexShard and torch_remat https://aitop.news/item/47532 AITOP en 2026-10-20T16:31:13.000Z Intelligence Index thresholds now far cheaper per task, Artificial Analysis finds https://aitop.news/item/47531 AITOP en 2026-10-20T16:30:02.000Z Shopify details continual learning loop with PyTorch at PyTorchCon 2026 https://aitop.news/item/47527 AITOP en 2026-10-20T16:19:42.000Z Mistral releases Voxtral 3 speech models, ranking high on benchmarks https://aitop.news/item/47522 AITOP en 2026-10-20T16:13:26.000Z AWS publishes X.509 PKI reference architecture for agent identity https://aitop.news/item/47520 AITOP en 2026-10-20T16:11:45.000Z SFResearch Study Tests Fragility of Memory-Based Self-Improving Agents https://aitop.news/item/47511 AITOP en 2026-10-20T15:55:16.000Z Bria Fibo 2.0 tops open weights image model leaderboards https://aitop.news/item/47509 AITOP en 2026-10-20T15:49:17.000Z Strategy Mosaic and Amazon Bedrock AgentCore build semantic layers for agents https://aitop.news/item/47507 AITOP en 2026-10-20T15:48:57.000Z AtScale MCP Server Connects Bedrock AgentCore Agents to Governed Data https://aitop.news/item/47485 AITOP en 2026-10-20T15:05:54.000Z Kaggle launches Human Behavior Bench for predicting human decisions https://aitop.news/item/47469 AITOP en 2026-10-20T14:13:56.000Z Qwen3.8-27B RL Post-Train Cuts Hallucination Rate in Half https://aitop.news/item/47462 AITOP en 2026-10-20T13:39:23.000Z SciResearchBench accepted to NeurIPS 2026