Quantara AI Daily · Wednesday, October 22, 2026AIDAILY
Quantara
Qianjue Technology Unveils Polibrain Predictive World Model
Qianjue Technology has unveiled Polibrain, a groundbreaking predictive world model leading global research in physical AI and embodied robotics.
模型发布/更新
新智元Qianjue Technology Debuts Polibrain Predictive World Model for Physical AI
Qianjue Technology announced Polibrain, an innovative predictive world model pioneering physics-grounded AI and robotic control.
新智元RoboColiseum Benchmark Evaluates Foundation Models in Embodied Robotics
The RoboColiseum benchmark assessed frontier models including GPT-6 Astra and Claude Opus 5.5, identifying key strengths in robotic spatial planning.
MemT 1.0 Integrates Native Working Memory into Transformer Architectures
MemT 1.0 introduces native persistent working memory branches into Transformer layers, enabling continuous learning and context consolidation.
MiniMax Releases M3.1 Flash with 70% Cost Reduction
MiniMax unveiled M3.1 Flash, delivering high performance alongside competitive API pricing of 1 RMB/M input tokens and 70% lower overall inference costs.
新智元Geek+ Gravity Control Model Boosts OOD Robotics Success Rate to 95%
Geek+ unveiled Gravity Control, using natural-language real-time corrective feedback to elevate out-of-distribution manipulation success from 5% to 95%.
Google ResearchFirst-PartyEmpowering frontline maternal care with AI-assisted ultrasound
The World Health Organization (WHO) estimates that more than 250,000 preventable deaths occur worldwide each year due to pregnancy-related complications. While prenatal ultrasound can improve outcomes and reduce fetal and maternal mortality, two-thirds of people globally lack acc
Learn more
Key takeaways State-of-the-art retrieval : Rerank 5 Pro leads on major public benchmarks, while Rerank 5 Fast delivers near-leading performance at lower cost. Strong across enterprise domains : Rerank 5 establishes itself as a leader across finance, healthcare, scientific researc
NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required
TL;DR NVIDIA Kumo Relational is a pre-trained relational foundation model that makes predictions directly on enterprise databases. Given a predictive query, it gathers up to 10,000 historical examples as context and predicts outcomes such as churn, demand, or fraud in a single fo
产品发布/更新
新智元DeepSeek Open-Sources DeepInfer High-Throughput Inference Engine
DeepSeek released DeepInfer, an open-source distributed inference framework delivering sub-millisecond latencies and deepseek-recipe codec tooling.
vLLM Team BlogAnnouncing Humming: Fast, Flexible Kernels for Quantized Inference
Ant Group and the vLLM community are open-sourcing Humming, a general-purpose, high-performance kernel library for quantized inference. Humming offers the following features: Broad device support : Supports NVIDIA GPU architectures from Turing onward, along with CUDA-compatible d
Latent Space[AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
Proudly brought to you by SerpAPI: Today’s release covers a LOT of ground, but you can scrape all this into markdown for your agents ! 50% token cost of JSON. We last covered the Inference Inflection in April , and historically most of the advances have been closed source decacor
Healthcare and Life Sciences Leaders Build on NVIDIA to Advance Drug Discovery and Expand Patient Care
News Summary: NVIDIA and the UW Medicine Institute for Protein Design debut RFD4-Proteina, a new open model that combines programmable protein design with search capabilities, unlocking new frontiers in biology. Siemens Healthineers is advancing the next generation of adaptive ra
Bringing the Spark to Europe: Local Intelligence Powers Next Wave of Agents
At GTC Berlin, NVIDIA is showcasing its full lineup of local AI hardware. RTX Spark , which launched last week , powers personal agents on Windows PCs; DGX Spark supports developers building and scaling applications on Linux; and DGX Station serves enterprise teams running fronti
Zijing Zhikang Debuts Multi-Agent Clinical Collaboration Architecture
Zijing Zhikang introduced a multi-agent hospital workflow platform, validating improved diagnostic precision across pilot partner hospitals.
NVIDIA Advances Isaac Platform for General-Purpose Robotics With New Reference Humanoid, Frameworks, Simulation Pipelines and Data
News Summary: Agile Robots announces Agile ONE SN, a humanoid reference platform built on NVIDIA Isaac GR00T and NVIDIA Jetson Thor with planned integration of the NVIDIA Halos safety stack, giving developers and researchers a complete platform for frontier humanoid development a
新智元HKUST Open-Sources Fourth University for Collaborative Human-Agent Research
HKUST released Fourth University, an open framework enabling autonomous scientific exploration and peer review between human researchers and AI agents.
Industry Dynamics
‘This Is Not a Drill’: Critical PHP Vulnerabilities Are Setting the Internet on Fire
Internet security and web administration teams around the world are scrambling to prevent widespread disaster. On Wednesday, the developers behind the much-used PHP programming language released emergency security updates designed to fix critical flaws that could be exploited to
PHP2Shell: AI uncovers critical RCE in PHP
Executive Summary Wiz Research uncovered multiple critical vulnerabilities in core PHP, a popular general-purpose scripting language that powers 70% of all websites in the world. Successful exploitation could allow attackers to take control of affected servers, steal sensitive da
AI 极客新闻eMudhra Intermediate CA Private Key Compromise and Mitigation
Cloudflare alerted the ecosystem to private key exposure across eight eMudhra intermediate certificates, publishing proactive revoking guidelines.
NVIDIA, Nokia and Leading Telecom Operators Turn Networks Into Distributed Intelligence Platforms
News Summary: Leading global telecom operators are expanding NVIDIA-powered computing across their networks to deliver advanced services closer to people and businesses, improve wireless experience and automate network operations. T-Mobile completes the world’s first AI-RAN call
NVIDIA and Supply Chain Partners Expand Capacity to Accelerate AI Infrastructure Buildout
New Capacity Agreements and Manufacturing Facilities to Support Volume Growth and Strengthen Supply Chain Resiliency News Summary: Substrate agreements with IBIDEN, Samsung Electro-Mechanics and Unimicron Technology Corporation will, in aggregate, drive a 2.5x increase in capacit
Research Papers
Solving a 31-year-old grand challenge in mathematics – the KLS conjecture
Microsoft researcher and collaborator combine new mathematical ideas with human-directed AI to resolve a central problem in high-dimensional geometry and probability At a glance Fundamental scientific breakthrough: Researchers resolved the Kannan-Lovász-Simonovits (KLS) conjectur
Evolving Agent Interfaces
Published on Oct 20 Authors: , , , , , , Abstract Agent S4 autonomously evolves computer-use interfaces from task experience, significantly improving open-source agent performance and reducing costs compared to larger proprietary models. Computer-use agents are limited by the int
Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition
Published on Oct 20 Authors: , , , , , , , , , , , , , , , , , , , , Abstract Sage-A1.5 is a 35B mixture-of-experts agentic model that improves long-horizon problem solving via adaptive task delegation, verification, and open scientific exploration. In this report, we introduce S
Large Flow Language Models
Published on Oct 20 Authors: , , , , , Abstract A large-scale continuous flow language model achieves efficient training, high-quality generation, and faster inference through blockwise flow design, self-speculative decoding, and multi-token prediction. Continuous flow language m
S-Agent 2.0: Spatial Reasoning with a General Coding Agent
Published on Oct 20 · Submitted by leoli on Oct 21 Authors: , , , , , , , , , , Abstract S-Agent 2.0 improves spatial reasoning by integrating a persistent scene memory and plain-language interpretation into a coding agent with 3D perception tools. Spatial reasoning asks a model
Matrix-Game 4.0: Harnessing Video Models for Real-Time Playable Game Worlds
Published on Oct 20 Authors: , , , , , , , , , , , , , , , , Abstract Matrix-Game 4.0 separates game logic from visual synthesis via an executable engine and autoregressive video renderer, achieving real-time interactive gameplay with low latency. Moving from controllable videos
World Models with Configurable Perception
Published on Oct 20 Authors: , , , Abstract ConfigWM uses task descriptions to steer visual embeddings into task-specific state spaces, filtering irrelevant noise for robust planning. Real-world observations are noisy and often contain excessive task-irrelevant information. Effec
Training the Whole Agent, Not Just the Model
Published on Oct 20 Authors: , , Abstract HarnessAdapt-RL jointly trains a coding policy and adapts its evaluation harness via mechanism-level skills and evolving tools, improving benchmark pass rates over fixed-harness reinforcement learning. Training coding agents typically upd
技巧与观点
a16z NewsWhy Hardware Specialization is Back
In 1990, Eugene Brooks of Lawrence Livermore told the Supercomputing conference that nothing would withstand the attack of the killer micros. “Micro” referred to microprocessor, meaning a general-purpose chip that did a good job at most types of computation. They were also being
金融科技
深潮 TechFlowAero Launches Multi-Chain Unified Liquidity Layer Across 7 Networks
Aero debuted as a unified liquidity fabric uniting Aerodrome and Velodrome AMM mechanisms across seven Layer-1 and Layer-2 blockchains.
AI趣闻
IT之家SpaceX Prepares Orbital Catch Attempt for Starship and Super Heavy Booster
SpaceX is slated to attempt complete mechanical tower recovery for both Starship and its Super Heavy booster in an upcoming test flight.
IT之家OpenAI Patches Critical PHP Flaws and Pledges $1B for Open-Source Security
OpenAI infrastructure security team helped identify critical PHP vulnerabilities while committing resources to protect underfunded open-source foundations.
Flashes
- GTC Berlin 2026: Live Updates From NVIDIA CEO Jensen Huang’s KeynoteNVIDIA Blog
- Google Cloud and NVIDIA expand co-engineering collaboration to scale infrastructure for the agentic eraGoogle Cloud AI
- NVIDIA Open Models Advance Europe’s Sovereign Intelligence Across IndustriesNVIDIA Blog
- Europe’s Industrial Leaders Reinvent Engineering and Manufacturing With NVIDIANVIDIA Blog
- NVIDIA Confidential Computing Expands, Enabling Frontier Model Builders to Securely Scale Across Global InfrastructureNVIDIA Blog
- Google Cloud and NVIDIA Announce General Availability of NVIDIA Vera Rubin InfrastructureNVIDIA Blog
- AI Runtime is Generally AvailableDatabricks Blog
- Mistral Advances AI for Engineering with NVIDIAMistral AI
- Advancing digital sovereignty with greater control and choiceMicrosoft Azure AI & Cloud
- European Mobility Leaders Build Safe Vehicles and Robotaxi Fleets on NVIDIA HyperionNVIDIA Newsroom
- NVIDIA Launches OpenShell Reference Architecture for Safe AI AgentsNVIDIA Newsroom
- European Ecosystem Builds World’s Most Efficient AI Factories With NVIDIA DSXNVIDIA Newsroom