Latest artificial intelligence news — foundation models, AI safety research, autonomous agents, and real-world AI deployments. Updated every 30 minutes from top AI publications.
Researchers behind the arXiv paper 2611.07354v1, titled MIKU: Material Invariance and Knowledge Unification for PBR Generation, proposed a unified framework that standardizes physically based rendering (PBR) asset repres…
Researchers posted arXiv paper 2611.07408v1 introducing PSGAgent, an agentic framework that lets multimodal large language models dynamically refine visual perception for panoptic scene graph generation through a think-a…
Nace AI released Drex, a family of four open-source decision models with 8B (DLM), 9B (v1.5), 35B with 3B active (v2.0), and 27B (Max) parameters that return a probability distribution over user-supplied options without …
A generative verifier called GenVer achieved the highest classification accuracy among evaluated baselines on four FactFB variants built from FB15k with perturbed negative triples, scoring between 88.56% and 91.52%, acco…
Researchers introduced TB-MathModel, a terminal-style benchmark that makes 18 real CUMCM and MCM contest problems from 2023–2025 agentic and verifiable, scoring agents' single JSON artifacts against reproductions of Outs…
Researchers introduced Agent-R$^2$, a real-source data synthesis framework that distills 4,000 task instances from 10,000 web-corpus candidates to train agents on realistic productivity work, according to the arXiv paper…
A preregistered mechanistic study of six open models from 14B to 675B parameters found that wrong sign tokens in multi-step LLM mathematics are committed late, by MLP layers in the last quarter of the network (70-100% pe…
A new paper (arXiv 2611.07071) proposes "mechanistic shaping," a framework that uses a model's internal representations to specify which concepts it should rely on and which it should express during training, rather than…
The IndiaAI Mission, run under MeitY, released three indigenous foundation models on November 3, 2026: Soket Labs' EKA-1, a 26.3B-parameter sparse hybrid model trained on roughly 7.5 trillion tokens, and ZenteiQ's Dhir 2…
Researchers present EDGAR, a diffusion-driven method for refining Gaussian-splatting head avatars built from monocular video or images, targeting artifacts such as blurring, stretching and temporal instability caused by …
A developer behind Pythos Oracle, a mathematics and physics AI system that pairs language models with deterministic math engines and independent verification, reported that the system solved all 50,000 problems in its 50…
Google has patented a transformer-based model called WebFormer that reads a web page's DOM tree structure alongside its text to extract fields such as title, price, brand and location across sites it has never seen. The …
Nvidia filed a patent application for a neural-network pipeline that estimates an object's shape, lighting and material from two photos taken under different lighting, such as with and without a flash. The four-network c…
Samsung has filed a patent for a system that trains a forecasting model to predict AI service response times and then uses gradient descent to find the smallest number of model instances that keeps responses within a man…
Google has filed a patent application describing a method for ranking which input data signals actually improve an AI model by hiding one signal at a time — replacing its values with a null marker — and measuring the per…
A developer's documentation-only build examining SAP Integration Suite's MCP Gateway found that when the ERP vendor also hosts the agent tools, tools are generated from existing API definitions and exposed via a Streamab…
Swarms founder Kye Gomez released Swarms v17 "InterLink" on November 2nd, 2026, adding six communication channels and a validated fan-out workflow that give multi-agent systems a shared conversation space instead of pass…
The Open AI Cellular (OAIC) project released OAIC-T v2.0.0, adding RANFusion, a simulator of 5G core network component behavior and performance, with InfluxDB support for storing network data and implemented KPI metrics.…
NVIDIA open-sourced the Cosmos Framework, the training and inference toolkit for its Cosmos 3 world model family, shipping it with five bundled AGENTS.md agent skills that let coding agents such as Claude Code, Codex CLI…
Open-weight models are the best-value option at just 3% of budgets on Village's LLM Atlas price-performance frontier as of October 31, 2026, down from 41% in June, with Google's Gemma 4 E4B and Xiaomi's MiMo-V2.6-Flash h…