https://aitop.news/item/54324 AITOP en 2026-11-05T20:44:39.000Z Stanford paper finds bits-per-byte is not tokenizer-agnostic https://aitop.news/item/54321 AITOP en 2026-11-05T20:42:06.000Z Hugging Face publishes 30k-word MoE training handbook https://aitop.news/item/54320 AITOP en 2026-11-05T20:41:56.000Z Hugging Face Publishes Comprehensive Guide to Training MoE Models https://aitop.news/item/54319 AITOP en 2026-11-05T20:41:45.000Z Hugging Face releases 70,000-word MoE training handbook https://aitop.news/item/54316 AITOP en 2026-11-05T20:38:50.000Z LangChain revamps async subagents in Deep Agents https://aitop.news/item/54303 AITOP en 2026-11-05T20:05:40.000Z Microsoft Research's Proof-Driven Development Kit Verifies AI-Generated Code https://aitop.news/item/54284 AITOP en 2026-11-05T19:32:07.000Z vLLM Ecosystem Powers Kimi K3 and GLM-5.2 on TPUs https://aitop.news/item/54283 AITOP en 2026-11-05T19:30:01.000Z LlamaIndex releases OCRia-1 document parsing model https://aitop.news/item/54282 AITOP en 2026-11-05T19:29:02.000Z Meta Rolls Out Muse Realtime Avatar in Muse App https://aitop.news/item/54280 AITOP en 2026-11-05T19:27:09.000Z webAI open-sources TwIL-8B, an 8.8B model built on Granite https://aitop.news/item/54275 AITOP en 2026-11-05T19:13:14.000Z StreamEval benchmark tests 38 models on real-time video AI https://aitop.news/item/54271 AITOP en 2026-11-05T19:08:04.000Z Science's Last Exam benchmarks AI research agents on reproducible work https://aitop.news/item/54268 AITOP en 2026-11-05T19:03:17.000Z METR reviews Anthropic's automated R&D risk claims https://aitop.news/item/54267 AITOP en 2026-11-05T19:02:45.000Z Tri Dao open-sources GPU kernel optimization agent https://aitop.news/item/54264 AITOP en 2026-11-05T19:02:07.000Z vLLM Ranks #2 in GitHub Octoverse Community Engagement https://aitop.news/item/54254 AITOP en 2026-11-05T18:50:16.000Z Verne Robotics V0.5 open-weight world action model https://aitop.news/item/54246 AITOP en 2026-11-05T18:37:49.000Z AI video market map 2026: 137 companies across 10 layers https://aitop.news/item/54237 AITOP en 2026-11-05T18:28:06.000Z OpenRouter reports 89% median agentic cache rate in 2026 https://aitop.news/item/54236 AITOP en 2026-11-05T18:27:48.000Z Google paper shows RAG poisoning works by echoing queries https://aitop.news/item/54232 AITOP en 2026-11-05T18:12:21.000Z Amazon Marketplace scores product page quality with SageMaker AI https://aitop.news/item/54218 AITOP en 2026-11-05T17:56:12.000Z ripple-65b preview from Ricursive launches on Together AI https://aitop.news/item/54217 AITOP en 2026-11-05T17:56:05.000Z swyx argues independent AA-* rebuilds are needed to catch benchmaxxing https://aitop.news/item/54214 AITOP en 2026-11-05T17:50:53.000Z Nanbeige4.3-33B open-weights model matches Qwen3.8 27B https://aitop.news/item/54209 AITOP en 2026-11-05T17:45:44.000Z OpenAI launches Private Safety Processing for API safety reviews https://aitop.news/item/54191 AITOP en 2026-11-05T17:33:46.000Z Replit rebuilds sandbox engine around microVMs for AI agents https://aitop.news/item/54180 AITOP en 2026-11-05T17:26:13.000Z NVIDIA Releases VFlash, a System-1 VLM for Physical AI https://aitop.news/item/54179 AITOP en 2026-11-05T17:24:30.000Z VFlash Parallel Vision Model Speeds Robot Inference 5.1x https://aitop.news/item/54174 AITOP en 2026-11-05T17:17:34.000Z Apache Spark 4.3 ships in Databricks Runtime 19 https://aitop.news/item/54141 AITOP en 2026-11-05T17:05:22.000Z Nous Research launches Hermes Agent Plugin Hackathon with $20,000 prizes https://aitop.news/item/54135 AITOP en 2026-11-05T17:03:37.000Z Nathan Lambert Adds Score-Centering to the RLHF Book https://aitop.news/item/54132 AITOP en 2026-11-05T17:02:55.000Z NVIDIA shows grouped execution for MoE optimization in PyTorch https://aitop.news/item/54130 AITOP en 2026-11-05T17:02:08.000Z Anyscale trains GLM-5.3-Flash at 1M-token context for genomics https://aitop.news/item/54112 AITOP en 2026-11-05T16:55:31.000Z Max Planck Paper Finds Reasoning Makes LLMs Share Mistakes https://aitop.news/item/54110 AITOP en 2026-11-05T16:54:55.000Z Stanford paper maps the in-context learning circuit in small LLMs https://aitop.news/item/54097 AITOP en 2026-11-05T16:52:41.000Z AutoSFT splits SFT data research across multi-agent roles https://aitop.news/item/54081 AITOP en 2026-11-05T16:49:10.000Z hunch builds a real-time learning local semantic decision system on Google's EmbeddingGemma 2 https://aitop.news/item/54078 AITOP en 2026-11-05T16:48:28.000Z Tencent Hunyuan Hy4 scores 35 on Artificial Analysis Index v5 https://aitop.news/item/54075 AITOP en 2026-11-05T16:47:57.000Z Together AI Launches Together Compiler for Polymorphic Inference https://aitop.news/item/54050 AITOP en 2026-11-05T16:37:47.000Z Mistral launches Le Chonk model and Studio managed agents https://aitop.news/item/54048 AITOP en 2026-11-05T16:37:23.000Z vLLM ships day-0 support for Baidu's ERNIE 5.5 Flash Base https://aitop.news/item/54040 AITOP en 2026-11-05T16:32:20.000Z MiMo-V2.6-Pro Tops AA-PersonalityBench in Question-Ending Responses https://aitop.news/item/54038 AITOP en 2026-11-05T16:29:20.000Z Baidu releases ERNIE 5.5 Flash Base with open weights https://aitop.news/item/54016 AITOP en 2026-11-05T16:13:15.000Z Bitkub MCP Package Re-Registration Exposes Agent Supply Chain Risk https://aitop.news/item/54003 AITOP en 2026-11-05T16:03:26.000Z Cognition Releases Flash Model Post-Trained on Inkling-small https://aitop.news/item/54001 AITOP en 2026-11-05T16:02:29.000Z Epoch AI Updates ECI With Bayesian Methodology https://aitop.news/item/53997 AITOP en 2026-11-05T15:47:47.000Z Recursa Paper Shows RSI Lifts Model Performance Without Fine-Tuning https://aitop.news/item/53973 AITOP en 2026-11-05T14:33:03.000Z CMU ServerBench Tests Coding Agents on LLM Serving Systems https://aitop.news/item/53953 AITOP en 2026-11-05T13:23:58.000Z Baidu open-sources ERNIE 5.5 Flash Base for continued training https://aitop.news/item/53927 AITOP en 2026-11-05T10:56:30.000Z Tencent Hunyuan Hy4 scores 35 on Intelligence Index v5 https://aitop.news/item/53855 AITOP en 2026-11-05T06:04:03.000Z Training Data and Routing Choices Push Open Models on ARC-AGI-2 https://aitop.news/item/53846 AITOP en 2026-11-05T05:59:39.000Z OpenAI brings GPT-Live outbound telephony API to GA https://aitop.news/item/53842 AITOP en 2026-11-05T05:30:32.000Z NVIDIA releases CUDA Toolkit 13.5 with Vera Rubin GA support https://aitop.news/item/53815 AITOP en 2026-11-05T04:12:49.000Z Systolic Linear Attention Optimises Gated Delta Attention for Google TPUs https://aitop.news/item/53754 AITOP en 2026-11-05T03:08:15.000Z Priors Beat Policies in LLM Test-Time Compute Allocation https://aitop.news/item/53723 AITOP en 2026-11-05T02:33:03.000Z Tsinghua and Berkeley open-source uni-robot-agent framework https://aitop.news/item/53711 AITOP en 2026-11-05T02:25:35.000Z DeepSeek Tokens Split Across 20 Inference Providers https://aitop.news/item/53709 AITOP en 2026-11-05T02:20:18.000Z OpenAI image API adds complexity parameter and SVG export https://aitop.news/item/53687 AITOP en 2026-11-05T01:07:26.000Z USTC releases open-source RNA foundation model structRFM in Nature https://aitop.news/item/53675 AITOP en 2026-11-05T00:26:58.000Z OpenAI model finds 1,000 Linux LPE vulnerabilities https://aitop.news/item/53672 AITOP en 2026-11-05T00:26:28.000Z OpenAI API adds marked-up images for targeted edits https://aitop.news/item/53671 AITOP en 2026-11-05T00:26:18.000Z Codex simplifies image editing with direct overlay edits https://aitop.news/item/53627 AITOP en 2026-11-04T21:24:11.000Z Stanford AI Lab Introduces Joules-to-GDP Framework for AI Energy Efficiency https://aitop.news/item/53624 AITOP en 2026-11-04T20:56:17.000Z Cognition's SWE-2 Flash Matches SWE-2 at One-Fifth the Price https://aitop.news/item/53621 AITOP en 2026-11-04T20:42:40.000Z Kimi K3.1 Pro lands on Databricks at launch https://aitop.news/item/53611 AITOP en 2026-11-04T20:22:23.000Z Anthropic uses Claude to map bacterial vault particle structure https://aitop.news/item/53608 AITOP en 2026-11-04T20:19:30.000Z SWE-2 Flash free for Devin Pro, Max and Teams users this month https://aitop.news/item/53607 AITOP en 2026-11-04T20:19:20.000Z Ripple-65B Preview: Stanford-affiliated Ricursive Intelligence releases 65B open-weight model https://aitop.news/item/53600 AITOP en 2026-11-04T19:49:46.000Z Ricursive Intelligence Releases Ripple-65B Preview Under Apache 2.0 https://aitop.news/item/53599 AITOP en 2026-11-04T19:43:27.000Z Developers build custom agentic CI on Temporal https://aitop.news/item/53587 AITOP en 2026-11-04T19:35:22.000Z PyTorch 2.15 Q&A Session on 5 November Release https://aitop.news/item/53585 AITOP en 2026-11-04T19:34:52.000Z Aether Launches Cloud Coding Agents With 19.4s Startup https://aitop.news/item/53578 AITOP en 2026-11-04T19:23:29.000Z NVIDIA Nemotron 3.5 Transcribe Tops Hugging Face Trending https://aitop.news/item/53558 AITOP en 2026-11-04T18:49:24.000Z Karpathy maps the inference chain from 1970s sci-fi to LLMs https://aitop.news/item/53544 AITOP en 2026-11-04T18:28:21.000Z LangChain demos agent booking dinner reservations via Managed Deep Agents https://aitop.news/item/53540 AITOP en 2026-11-04T18:26:23.000Z Gemini API CLI Workflow for Testing Multimodal Voice Features https://aitop.news/item/53535 AITOP en 2026-11-04T18:23:11.000Z Agent Arena data shows 66% cost gap in hour-long agent tasks https://aitop.news/item/53533 AITOP en 2026-11-04T18:22:50.000Z Axiom releases 16M-line Lean formalization of the Stacks Project https://aitop.news/item/53532 AITOP en 2026-11-04T18:22:32.000Z Gemini 4 Carbon leads cost scaling in Agent Arena data https://aitop.news/item/53529 AITOP en 2026-11-04T18:18:56.000Z Google Research Introduces Agile Deliberation Framework for VLMs https://aitop.news/item/53527 AITOP en 2026-11-04T18:10:02.000Z Axiom Formalizes Stacks Project in Lean Using AI Agents https://aitop.news/item/53523 AITOP en 2026-11-04T18:01:52.000Z Anyscale session on LLM post-training, RLVR and SkyRL https://aitop.news/item/53521 AITOP en 2026-11-04T17:56:17.000Z Axiom Math formalizes Stacks Project in Lean with agent swarm https://aitop.news/item/53520 AITOP en 2026-11-04T17:56:11.000Z Atomic Local AI Server runs local models via multiple engines https://aitop.news/item/53519 AITOP en 2026-11-04T17:45:36.000Z Together Compiler Optimizes Inference for Production Traffic https://aitop.news/item/53518 AITOP en 2026-11-04T17:39:32.000Z Artificial Analysis Makes Time per Task Its Headline Benchmark Metric https://aitop.news/item/53508 AITOP en 2026-11-04T17:30:49.000Z David Pawlan on Assistant Benchmark and 147 agents https://aitop.news/item/53504 AITOP en 2026-11-04T17:22:53.000Z Kimi K3.1 Pro rolls out on Ollama Cloud with zero data retention https://aitop.news/item/53499 AITOP en 2026-11-04T17:19:05.000Z RSI-Craft paper trains models to run their own post-training https://aitop.news/item/53487 AITOP en 2026-11-04T17:13:05.000Z Zyphra's ZHA hierarchical attention speeds up long-context models https://aitop.news/item/53438 AITOP en 2026-11-04T16:51:01.000Z Tsinghua paper shows AI memory deletion fails 98.5% of forget requests https://aitop.news/item/53422 AITOP en 2026-11-04T16:41:52.000Z NVIDIA Open-Sources Auto Ontology for Enterprise Data Semantics https://aitop.news/item/53420 AITOP en 2026-11-04T16:38:49.000Z Fish Audio Drama 3 voice model coming soon to Alibaba Cloud https://aitop.news/item/53419 AITOP en 2026-11-04T16:38:42.000Z Alibaba Tair team co-leads SGLang agentic cache scheduling roadmap https://aitop.news/item/53402 AITOP en 2026-11-04T16:28:24.000Z Hugging Face and partners launch RoboColosseum open robot eval initiative https://aitop.news/item/53383 AITOP en 2026-11-04T16:14:00.000Z milvus-backup 0.6 adds snapshot restores with indexes on Milvus 3.0.1 https://aitop.news/item/53365 AITOP en 2026-11-04T16:02:41.000Z MolmoAct 2 ranks second to GPT-6 Astra in RoboColosseum robot tests https://aitop.news/item/53364 AITOP en 2026-11-04T16:02:23.000Z NVIDIA adds 1/2 DPO training to NeMo Megatron Bridge https://aitop.news/item/53354 AITOP en 2026-11-04T15:21:02.000Z Cohere Embed 5 offers Pro and Fast models in one space https://aitop.news/item/53353 AITOP en 2026-11-04T15:20:57.000Z Cohere Embed 5 and Rerank 5 launch on OpenRouter https://aitop.news/item/53342 AITOP en 2026-11-04T14:19:32.000Z MIT 6.S950 Course Notes Cover Outsourced Data Risks in AI Agents https://aitop.news/item/53341 AITOP en 2026-11-04T14:19:20.000Z TemPad Dev connects Figma to coding agents via MCP https://aitop.news/item/53330 AITOP en 2026-11-04T13:01:54.000Z NVIDIA Paper Shows Aggressive Context Compaction Matches Full-History Agents https://aitop.news/item/53305 AITOP en 2026-11-04T11:13:22.000Z Mistral Large 4 agents find and patch a Django CVE https://aitop.news/item/53303 AITOP en 2026-11-04T11:01:22.000Z Multimodal tool-augmented agent targets AMD follow-up care https://aitop.news/item/53261 AITOP en 2026-11-04T08:09:43.000Z Tencent Hunyuan Hy4 Ranks #14 on Agent Arena https://aitop.news/item/53248 AITOP en 2026-11-04T07:21:44.000Z Tencent Hunyuan Hy4 launches with vision capabilities https://aitop.news/item/53228 AITOP en 2026-11-04T06:04:17.000Z Adversarial Speculative Reasoning Boosts Long-Horizon Agent Accuracy https://aitop.news/item/53168 AITOP en 2026-11-04T04:00:46.000Z MobileSpec Trains LLM Specialists on iPhones Using Forward Passes Only https://aitop.news/item/53146 AITOP en 2026-11-04T03:52:06.000Z LeanSparse Cuts GPU Memory for Sparse-Attention LLM Serving https://aitop.news/item/53133 AITOP en 2026-11-04T03:06:32.000Z Playwright clarifies it bundles no browsers amid security misconception https://aitop.news/item/53110 AITOP en 2026-11-04T02:52:41.000Z TNAS Diagnostic Exposes Silent Action Attacks on Reasoning VLA Models https://aitop.news/item/53100 AITOP en 2026-11-04T02:50:21.000Z Multi-layer Indexer Speeds DeepSeek Sparse Attention Decode https://aitop.news/item/53093 AITOP en 2026-11-04T02:49:14.000Z GC-MoE Operator Accelerates MoE Inference with CUDA Green Contexts https://aitop.news/item/53086 AITOP en 2026-11-04T02:27:19.000Z Skill++ keeps agent skill routers accurate as libraries scale https://aitop.news/item/53048 AITOP en 2026-11-04T02:19:08.000Z LlamaIndex speeds up OCRia-1 token generation 10× on a B200 https://aitop.news/item/53047 AITOP en 2026-11-04T02:19:04.000Z LlamaIndex splits benchmarks into frontier and specialized categories https://aitop.news/item/53040 AITOP en 2026-11-04T02:15:45.000Z ParetoSaddler optimizes agent harnesses for quality and cost https://aitop.news/item/53016 AITOP en 2026-11-04T02:05:08.000Z RoboArchitect rewrites robot control code automatically https://aitop.news/item/53013 AITOP en 2026-11-04T02:04:38.000Z Harvard and MIT paper traces reasoning control failures to "Okay" openers https://aitop.news/item/53006 AITOP en 2026-11-04T02:04:29.000Z Artificial Analysis adds GPU rental pricing to AA-AgentPerf benchmark https://aitop.news/item/52978 AITOP en 2026-11-04T02:00:09.000Z vLLM v0.33.0 Ships PyTorch 2.14, New Models and NVIDIA Gains https://aitop.news/item/52969 AITOP en 2026-11-04T01:52:21.000Z Pika H3 Flex gets community Turbo LoRA within 24 hours https://aitop.news/item/52967 AITOP en 2026-11-04T01:47:22.000Z AA-PersonalityBench: open weights models lag proprietary on truthfulness https://aitop.news/item/52965 AITOP en 2026-11-04T01:47:03.000Z Artificial Analysis Updates Six Capability Indexes to v2.0 https://aitop.news/item/52962 AITOP en 2026-11-04T01:46:32.000Z Anthropic Open-Sources FlashProve for Zero-Knowledge LLM Inference Verification