https://aitop.news/item/50781 AITOP en 2026-10-28T18:43:58.000Z AWS ships Ray Train deep learning containers for EFA training on EKS https://aitop.news/item/50780 AITOP en 2026-10-28T18:43:51.000Z Amazon Bedrock AgentCore Reference Blueprint for Governed Agentic AI Platforms https://aitop.news/item/50779 AITOP en 2026-10-28T18:41:12.000Z Muse Spark Open Scores on ARC-AGI Benchmarks Revealed https://aitop.news/item/50776 AITOP en 2026-10-28T18:38:14.000Z Google DeepMind's CALLIOPE maps LLM topics under differential privacy https://aitop.news/item/50761 AITOP en 2026-10-28T17:52:47.000Z PyTorch 2.15 ships with 3,117 commits and NCCL rewrite https://aitop.news/item/50741 AITOP en 2026-10-28T17:09:32.000Z DeepSeek V4.1 Flash tops FrontierCode 2.0 among open-weight models https://aitop.news/item/50737 AITOP en 2026-10-28T16:58:46.000Z AssemblyAI-TTS-1 Debuts in Artificial Analysis Voice Arenas https://aitop.news/item/50733 AITOP en 2026-10-28T16:40:49.000Z Artificial Analysis Launches AA-Video-I2V v2.1 Benchmark https://aitop.news/item/50723 AITOP en 2026-10-28T16:10:49.000Z ARC-AGI scores follow a power curve across benchmarks https://aitop.news/item/50711 AITOP en 2026-10-28T15:58:10.000Z Legora joins Claude for Legal with in-Claude agent integration https://aitop.news/item/50699 AITOP en 2026-10-28T15:32:47.000Z NVIDIA Nemotron 3.5 Transcribe speech model launches on vLLM https://aitop.news/item/50695 AITOP en 2026-10-28T15:10:44.000Z NVIDIA Nemotron 3.5 Transcribe ties for #1 open-weight STT https://aitop.news/item/50687 AITOP en 2026-10-28T15:00:53.000Z NVIDIA MotionBricks generates character motion at 15,000 fps https://aitop.news/item/50673 AITOP en 2026-10-28T14:05:56.000Z MiniCPM-V 4.7 runs 35B MoE on phones via LES https://aitop.news/item/50671 AITOP en 2026-10-28T13:51:09.000Z Google releases EmbeddingGemma 2, a 744M multimodal embedding model https://aitop.news/item/50661 AITOP en 2026-10-28T13:15:03.000Z Vela 3.0 Router from vLLM, KR Labs and AMD https://aitop.news/item/50658 AITOP en 2026-10-28T13:05:50.000Z MiniCPM5-2B-Jev turns MiniCPM5-2B into a decision model https://aitop.news/item/50610 AITOP en 2026-10-28T09:07:14.000Z Faraday Technology lands 2nm AI chip design win on Samsung process https://aitop.news/item/50587 AITOP en 2026-10-28T07:04:52.000Z MiniCPM-OCR pairs recognition with a checker model https://aitop.news/item/50566 AITOP en 2026-10-28T06:07:40.000Z OM-1 Humanoid Learns Manipulation from Sensorized-Glove Data https://aitop.news/item/50539 AITOP en 2026-10-28T03:16:44.000Z MiniCPM-OCR suite: open-source OCR models from ModelBest https://aitop.news/item/50529 AITOP en 2026-10-28T02:55:22.000Z SynthAPI Pipeline Derives Scaling Laws for Tool-Use Agents https://aitop.news/item/50520 AITOP en 2026-10-28T02:50:31.000Z Theory Explains Why Unsupervised RLVR Collapses on Large Datasets https://aitop.news/item/50516 AITOP en 2026-10-28T02:47:55.000Z Study Explains Why DPO Updates Stay Sparse via NTK Theory https://aitop.news/item/50437 AITOP en 2026-10-28T01:30:20.000Z GateRQ Quantizes DeepSeek Engram Memory Tables by Gate Weight https://aitop.news/item/50401 AITOP en 2026-10-28T00:09:23.000Z Spark-X2.6-1.7B on-device agent model ships with vLLM support https://aitop.news/item/50387 AITOP en 2026-10-27T23:26:55.000Z Drex for Search tops Snowflake's MADQA leaderboard at 94.7% https://aitop.news/item/50370 AITOP en 2026-10-27T22:27:11.000Z Together GPU Clusters Automate Node Repair from GPU Fault to Healthy https://aitop.news/item/50369 AITOP en 2026-10-27T22:11:06.000Z Cartesia Managed Agents hits Elo 1115 on Speech Agent Arena https://aitop.news/item/50365 AITOP en 2026-10-27T21:58:13.000Z Lean-Verified Tokenizers Target Training-Inference Mismatch in RL Pipelines https://aitop.news/item/50332 AITOP en 2026-10-27T19:58:45.000Z OpenBMB releases MiniCPM-OCR 1.0, a 1.3B open-source OCR model https://aitop.news/item/50321 AITOP en 2026-10-27T19:04:47.000Z Microsoft Research and Georgia Tech release MANDATE agent failure taxonomy https://aitop.news/item/50319 AITOP en 2026-10-27T18:59:30.000Z Ling 3.1 Flash 1019 open weights model raises AA index to 44 https://aitop.news/item/50308 AITOP en 2026-10-27T18:34:01.000Z LMSYS releases miniTE for faster BF16 training https://aitop.news/item/50297 AITOP en 2026-10-27T17:57:18.000Z Liquid AI launches LFM3 open-weight model family https://aitop.news/item/50293 AITOP en 2026-10-27T17:47:08.000Z Google Research Details Sleep-Phase Method for Continual LLM Learning https://aitop.news/item/50285 AITOP en 2026-10-27T17:18:39.000Z Fix evaluation bias to unlock better models on Amazon Bedrock https://aitop.news/item/50271 AITOP en 2026-10-27T16:53:44.000Z MiniCPM-OCR 1.0 and Jev: open-source OCR parsing and verification models https://aitop.news/item/50270 AITOP en 2026-10-27T16:52:10.000Z SoMark open-sources OXR-1.0 document parsing model https://aitop.news/item/50265 AITOP en 2026-10-27T16:41:45.000Z AI2 Eval Finds Frontier LLMs Inconsistent at Judging Novelty https://aitop.news/item/50257 AITOP en 2026-10-27T16:16:59.000Z HASP Converts Agent Failures into Reusable Skill Programs https://aitop.news/item/50242 AITOP en 2026-10-27T15:28:58.000Z Microsoft MAI Image 2.7 Preview climbs to No.3 on LMArena https://aitop.news/item/50239 AITOP en 2026-10-27T15:20:57.000Z Alibaba and Cornell paper benchmarks verification designs for coding agents https://aitop.news/item/50234 AITOP en 2026-10-27T15:10:52.000Z Microsoft's MAI-Image-2.7-Preview Ranks #2 on AA-Image-Editing v2.1 https://aitop.news/item/50211 AITOP en 2026-10-27T14:04:48.000Z CoreWeave expands Global AI Network to Singapore, Indonesia, UAE https://aitop.news/item/50205 AITOP en 2026-10-27T13:59:36.000Z You.com, NVIDIA and CoreWeave boost Nemotron 3.5 Lightning BrowseComp score https://aitop.news/item/50199 AITOP en 2026-10-27T13:58:07.000Z Running π0.5 Embodied AI on a Single RTX 4070 Ti SUPER https://aitop.news/item/50190 AITOP en 2026-10-27T13:14:46.000Z BAAI's AREX-2 (27B) Sets Sub-40B SOTA on BrowseComp and HLE https://aitop.news/item/50177 AITOP en 2026-10-27T12:26:52.000Z Hi3D open-sources v1.0 image-to-3D model https://aitop.news/item/50169 AITOP en 2026-10-27T11:34:47.000Z Google DeepMind researchers develop AudioFM audio representation framework https://aitop.news/item/50167 AITOP en 2026-10-27T11:23:06.000Z ForgeKernel from ModelBest ranks #1 on CANNBench and FlashInfer-Bench https://aitop.news/item/50151 AITOP en 2026-10-27T10:26:47.000Z Jev Matches Frontier LLM Accuracy in RAG Claim Verification https://aitop.news/item/50136 AITOP en 2026-10-27T09:26:50.000Z GenXR-Bench shows coding AI fails at 3D spatial XR tasks https://aitop.news/item/50132 AITOP en 2026-10-27T09:06:00.000Z 文心 Unlimited OCR 2 processes 3,623-page Harry Potter in 90 minutes https://aitop.news/item/50112 AITOP en 2026-10-27T07:46:49.000Z Baidu releases Unlimited-OCR-2 for whole-book OCR https://aitop.news/item/50090 AITOP en 2026-10-27T05:47:06.000Z Katamari strategy: applied AI firms training their own models https://aitop.news/item/50084 AITOP en 2026-10-27T05:00:47.000Z Claude and ChatGPT cut detailed reasoning traces https://aitop.news/item/50043 AITOP en 2026-10-27T04:23:50.000Z Hamel Husain urges skepticism toward AI benchmark results https://aitop.news/item/50015 AITOP en 2026-10-27T03:26:22.000Z d3-ultra multimodal model released with Apache-2.0 weights https://aitop.news/item/49977 AITOP en 2026-10-27T02:42:54.000Z Cachelint Detects Silent Prompt-Caching Bugs in LLM Agent Systems https://aitop.news/item/49946 AITOP en 2026-10-27T02:24:37.000Z Speculative Decoding for MoE Models: Deduplication and Draft Length https://aitop.news/item/49917 AITOP en 2026-10-27T01:38:00.000Z AutoTavern: LLM Agents Build Hearthstone Battlegrounds AI Rivaling Top Players https://aitop.news/item/49906 AITOP en 2026-10-27T01:20:58.000Z ProtAgent-8B Matches Frontier Models on Protein Tool Tasks https://aitop.news/item/49885 AITOP en 2026-10-27T00:44:17.000Z OpenAI releases GPT-6.1 Terra with visual reasoning focus https://aitop.news/item/49857 AITOP en 2026-10-26T22:23:40.000Z Artificial Analysis launches Code2Video-AA video benchmark