Knowledge catalogue
Google DeepMind: Google DeepMind introduces Gemini Robotics-ER 1.6 robotic reasoning model, says it shows significant spatial and physical reasoning improvements over ER 1.5 — For robots to be truly h
VAKRA (Versatile Agent and Knowledge Reasoning Assessment) is a benchmark developed by IBM Research to evaluate AI agents on complex reasoning, tool use, and multi-step task execution across diverse d
IBM Research has demonstrated that mid-training — a dedicated training phase between initial pre-training and fine-tuning — is essential for improving reasoning capabilities in large language models (
NV‑Reason‑CT is a purpose‑built vision‑language model for volumetric CT that uses a full 3‑D Vision Transformer encoder coupled with a Qwen 3.5‑4B language model to generate radiologist‑style chain‑of
Stealing Reasoning Traces from Proprietary LLM APIs A vanity domain name (stolen-thoughts.com) for a neat paper: Anthropic, OpenAI, and Google return encrypted chain-of-thought blocks to clients that
Will Knight / Wired: Researchers find that feeding a frontier model's encrypted reasoning traces to a weaker model from the same provider can make it output the traces in plaintext — Researchers devis
A path, a fence, a knot. MindTopo sets a new benchmark for testing how AI understands topological relationships and highlights new opportunities to strengthen spatial reasoning and planning. The post
Alibaba released the open‑weights Qwen3.8‑2.4T‑A95B (Qwen3.8‑Max), a fine‑grained mixture‑of‑experts model with 2.4 trillion parameters, hybrid full‑ and linear‑attention, a one‑million‑token context
I released LLM 0.32 this morning, the most significant new version of LLM since the initial launch of the project. The new version includes support for visible reasoning traces, server-side provider t
NVIDIA Alpamayo 2 Super is a publicly available 34‑billion‑parameter vision–language–action model that merges a 32‑B Cosmos 3 Super Reasoner with a 2‑B action‑expert diffusion network. It produces uni
Large language models can exhibit emergent reasoning behaviors, often manifested as recurring lexical patterns (e.g., “wait,” indicating verification). However, complex reasoning trajectories remain s
Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM’s autoregressive decoding may limit the ability to revisit and refine earlier t
NVIDIA Nemotron 3 Nano Omni is a 30B hybrid mixture-of-experts model that brings multimodal perception and reasoning into a single system, natively supporting text, image, video, and audio inputs whil
The NVIDIA Nemotron Model Reasoning Challenge on Kaggle attracted over 5,000 participants who all began from the same open model, benchmark, and infrastructure. The strongest solutions treated reasoni
Learn how to run the full Amazon Bedrock Automated Reasoning policy lifecycle from your coding agent. A suite of open source Agent Skills builds, reviews, tests, debugs, deploys, and validates a custo
Traditional APM can collect every span and still leave developers guessing about intent, causality, and drift. As agents multiply, the observability stack must learn to interpret the systems it watche
Amazon Bedrock now supports automatic Automated Reasoning policy refinement. The refinement engine diagnoses failing tests and proposes formal-logic fixes for rule issues and language issues, and you
In this post, we explore an idea for generating thinking tokens for datasets that lack reasoning traces in SFT customization. We first examine the reasoning suppression problem, then introduce Self-Di
Modern Large Language Models achieve impressive reasoning capabilities with long Chain of Thoughts, but they incur substantial computational cost during inference, and this motivates techniques to imp
Databricks explores how agentic AI systems can reason over both structured data (such as tables and databases) and unstructured data (such as documents and text) to answer complex business questions.
What is spatial reasoning? AI is moving beyond the digital world. The next generation of AI systems needs to understand and interact with physical environments, from robots to autonomous systems navig
Mike Wheatley / SiliconANGLE: At Dreamforce, Salesforce announces Koa, a specialized CRM reasoning model built on Nvidia's Nemotron architecture, designed to power AI agents in Agentforce — Salesforce
Google LLC is trying to address the latency problem associated with voice-based artificial intelligence agents with the launch of Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking today. They’re b
GPT‑6 Astra is OpenAI’s latest flagship model, outperforming GPT‑5.6 across virtually all common benchmarks—most noticeably in math, coding, and especially 3‑D rendering/animation tasks. The architect
Gemini Robotics-ER 1.6, introduced by Google DeepMind, is a significant upgrade to their reasoning-first robotics model that specializes in visual and spatial understanding, task planning, and success
Semafor: Reflection unveils Beam, an open model it says rivals GLM 5.2 on reasoning while using 3x-4× less compute and approaches Qwen3.8-Max on coding and agentic tasks — Reflection AI, the startup b
IBM has released Granite 4.2, an open‑source agentic AI model that adds native reasoning, tool use, coding instruction following and speech capabilities to enterprise agents. The suite of pre‑built mo
Businesses run on fast decisions, but the teams who hold the answers are often buried under a backlog of routine requests, leaving users waiting in line for insights they need now. Today, we are bring
NVIDIA Cosmos 3 is a frontier foundation model for physical AI that combines physical reasoning, world generation, and action generation within a single open model. The model uses a Mixture-of-Transfo
Tom Warren / The Verge: Sources: at Build, Microsoft plans to unveil a Copilot “super app”, a new reasoning AI model developed by Microsoft AI, and lots of Windows developer features — Build will incl
NVIDIA Cosmos 3 is an open-source omni-model designed for physical AI reasoning and action tasks, representing an advancement in multimodal AI systems. The model integrates multiple modalities to enab
Dean Takahashi / GamesBeat: General Intuition, which trains AI agents in spatial reasoning via gameplay footage, raised 220M at a 6.2B valuation, for 650M+ in total funding — General Intuition has rai
Tae Kim / Key Context: Astra working with Blender via computer use feels like magic, showing computer use could be the fourth demand wave after chatbots, reasoning, and agentic coding — AI computer to
NVIDIA Jetson can run compact open‑source reasoning models such as Nemotron 3.5 Lightning (a mixture‑of‑experts architecture with 30 B total parameters but only ~3 B active per token) and dense Qwen 3
Celia Ford / Transformer: OpenAI says it can't read all of Astra's reasoning and admits covert sandbagging would likely go uncaught, yet still calls it the world's most aligned model — OpenAI is haili
The Information: Source: OpenAI's Astra model uses “recurrent depth”, a technique that improves cost and performance but obscures the AI's reasoning, making it harder to monitor — OpenAI says its fort
SemiAnalysis: With each successive era of AI, from early scaling, to reasoning, to agentic, open models have taken half as long to catch up to the first closed model — Comparing open vs. closed models
Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposit
David Gewirtz / ZDNET: Opus 5 improves coding, reasoning efficiency, and prompt-cache-friendly tool use, and is priced at 5/1M input tokens and 25/1M output, same as Opus 4.8 — ZDNET's key takeaways —
OpenAI: GPT-5.6 Sol matches Mythos Preview on ExploitBench, adds Ultra mode with subagents for complex workflows, and max reasoning for deep problem-solving — We're beginning a limited preview of the
Rebecca Bellan / TechCrunch: General Intuition, which trains AI agents in spatial reasoning via gameplay footage, raised 320M led by Khosla at a 2.3B valuation, for $454M in total funding — As soon as
Ivan Mehta / TechCrunch: Google upgrades NotebookLM, which now runs on Gemini 3.5 and Antigravity, to deliver new agentic capabilities and more advanced reasoning for AI Ultra users — Google on Monday
NVIDIA's Nemotron 3 Ultra is a 550B-parameter Mixture-of-Experts model with 55B active parameters, optimized for orchestrating complex, long-running agent workflows by combining frontier reasoning and
Engadget: A live blog of Microsoft Build 2026, where the company is set to unveil a Copilot “super app”, a new reasoning model, and Windows improvements for developers — We're expecting the presentati
Jay Peters / The Verge: Microsoft debuts MAI-Thinking-1, its first advanced reasoning AI model, trained “from the ground up on clean data, without distillation from third-party models” — MAI-Thinking-
Enterprise decision intelligence startup RelationalAI Inc. is advancing its capabilities for Snowflake Inc.’s AI Data Cloud platform. At Snowflake Summit 2026 today, it announced a series of updates t
Microsoft announced a bunch of new in-house AI models at Build 2026, including a new 'flagship' model: MAI-Thinking-1. It's an ambitious step into model development for Microsoft, which introduced its
Carl Franzen / VentureBeat: xAI launches Grok 4.3, featuring “always-on reasoning”, 1M token context window, and low API pricing, and releases a voice cloning suite called Custom Voices — While Elon M
Anthropic: Anthropic says it has fixed three causes of recent Claude Code quality issues: reduced default reasoning, a caching bug, and a system prompt to reduce verbosity — We traced recent reports o
Madison Mills / Axios: OpenAI says GPT-5.5's improvements are strongest in agentic coding, computer use, and early scientific research, which require reasoning across longer contexts — OpenAI on Thurs
This Google Research work presents guidelines for synthetic data mechanism design and provides insights into generating and evaluating synthetic data at scale. The research introduces a reasoning-driv
In this post, you'll learn why probabilistic AI validation falls short in regulated industries and how Automated Reasoning checks use formal verification to deliver mathematically proven results. You'
Anthropic PBC today opened access to Claude Opus 4.7, the latest addition to its popular line of large language models. The company says that the LLM is significantly better than its predecessor at co
Exposure management company Tenable Holdings Inc. today unveiled new capabilities for Hexa AI, the agentic engine inside its Tenable One Exposure Management Platform, pushing the tool into general ava
Just three weeks after its last large language model release, Google LLC today launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber. The two LLMs are based on the same technical foundation. The primar
SpaceXAI today released Grok 4.6, a large language model that it says can outperform Anthropic PBC’s Claude Fable 5 in some areas. SpaceXAI was known as xAI until last month. The Elon Musk-founded art
Retrieval-augmented generation (RAG) enhances large language models (LLMs) with external knowledge but still suffers from long contexts and disjoint retrieval–generation optimization. In this work, we
AI agents on foundation models often misapply healthcare and life sciences decision frameworks, citing the right guideline but applying it incorrectly. This post shares 38 open-source agent skills acr
Every few decades, someone announces that science has reached its end. In 1903, the revered physicist Albert Michelson wrote that the “facts of physical science have all been discovered.” In the 1980s
AcuRite is discontinuing its My AcuRite app, citing obsolete technology and the unsustainability of maintaining two separate apps. The company stated that ongoing maintenance, security updates, and co