The daily radar on models, frameworks, and hardware to run AI locally. LLMs, LangChain, Chroma, mini-PCs, and everything you need for a distributed "in-house" brain.
AI-generated articles from feeds, with space for human editorial layer above the raw content.
First Linux benchmarks of the Intel Arc B390 GPU, integrated in high-end Panther Lake models. The Xe3 graphics card, equipped with 12 Xe cores, promises interesting performance in desktop and mobile environments for graphics and compute workloads.
ASRock has issued a statement regarding a new round of Ryzen 9000 CPU failures. The company says it is actively collaborating with AMD to identify and resolve the cause of the problem, which appears to affect a limited number of units.
Bound, a London-based automated FX risk management platform, has closed a $24.5 million Series A funding round. The funds will be used to expand into Europe and develop new perpetual hedging products, amid increasing volatility in currency markets.
OpenAI introduces Frontier, an enterprise platform designed for building, deploying, and managing AI agents. Frontier offers features such as shared context, onboarding, permission management, and centralized governance.
Canonical and SpacemiT announced that Ubuntu Linux will be officially supported on SpacemiT's new K3 RISC-V SoC. What makes the K3 interesting is being one of the first available RISC-V RVA23 designs.
Reports of access issues to the Hugging Face platform have surfaced online. Some users report being unable to access the platform, while others claim that core services remain operational. The cause and extent of the problem are not yet clear.
Tesla's large-scale production of Optimus robots heavily relies on the Chinese supply chain. The article highlights how trade tensions between the United States and China could pose a significant risk to Tesla's robotics ambitions.
The vLLM team introduced vLLM-Omni, a system designed for any-to-any multimodal models handling text, images, video, and audio. The architecture includes stage-based graph decomposition, per-stage batching, and flexible GPU allocation, achieving up to 91.4% JCT reduction tested with Qwen-Image-2512.
Epic Games is completely redesigning its launcher, aiming to make it lighter, more stable, and rich in social features. The mid-year update will include private DMs, customizable player profiles, and independent live chats, improving the overall user experience.
Lawhive, a UK-founded legaltech startup focused on democratizing legal services, has raised $60 million in Series B funding. The goal is to accelerate its US expansion, where it offers an AI platform to simplify everyday legal processes, reducing time and costs for both clients and lawyers.
Multiple newly disclosed bugs in the popular workflow automation tool n8n could allow attackers to hijack servers, steal credentials, and quietly disrupt AI-driven business processes. The patch meant to close a severe expression bug fails to stop attackers with workflow access.
OpenNebula highlights how data sovereignty is becoming an increasing concern for private companies, not just the public sector. Policies, licensing, and costs influence decisions, pushing towards greater control over data location and management.
The first beta release of Krita 6.0 is now available, a featureful digital painting program, re-based against the Qt6 toolkit. Krita 5.3 Beta is also being released for those sticking to Qt5. The update introduces improvements in color management and Wayland support.
A Reddit user is seeking alternatives to ChatGPT's Deep Research for running in-depth analysis with local LLMs. Their current setup includes 3x 3090 GPUs, OpenWebUI, and SearXNG, but the accuracy isn't comparable to ChatGPT. The article explores potential solutions and hardware configurations to enhance local research capabilities.
DeFi platform Step Finance suffered a $40 million breach. Hackers compromised executives' devices to gain illicit access to funds. The attack highlights the importance of device security in the crypto and DeFi space.
For those not fond of the increasing use of the Rust programming language within the Linux kernel or FreeBSD's considerations for Rust in its kernel, you can perhaps find refuge within NetBSD. One of the NetBSD developers has explained why you likely won't be finding Rust code within the NetBSD kernel anytime soon.
Microsoft researchers have unveiled a scanning method to identify poisoned AI models with backdoors, even without knowing the specific trigger or the attack's ultimate goal. The method exploits the tendency of these models to memorize training data and exhibit internal anomalies when processing the malicious trigger. This capability fills a security gap in open-source models, often reused to reduce training costs.
Entertainment about or made with artificial intelligence has been missing the mark with viewers over the past year. After a period of high interest, audiences may be showing signs of fatigue towards this type of content.
Biotech startup QT Sense has secured €4 million to accelerate its Quantum Nuova platform, a technology that lets scientists observe cellular processes in real time and reveal biochemical activity linked to disease. The funding includes seed investments and contributions from research programs.
A graph produced by METR, an AI research nonprofit, has become a benchmark for evaluating the progress of large language models (LLMs). However, its interpretation is often a source of confusion. The analysis primarily focuses on coding tasks and measures the time it takes humans to complete tasks that AI can successfully perform, not the duration of the models' autonomy. Despite the limitations, the study offers a concrete metric for evaluating the evolution of AI.
Nvidia is reportedly seeking faster deliveries of HBM4 memory from Samsung, amid a global crunch in high-bandwidth memory supply. The move highlights the competition to secure resources for upcoming AI accelerators.
Samsung is strengthening cybersecurity measures in its semiconductor supply chain to prevent leaks of sensitive technological information. The initiative aims to protect intellectual property and trade secrets in the chip industry.
A report by Invest Europe and EIF analyzes the European venture capital landscape, highlighting how gender, education, and geographical location influence startup funding. The report shows that VC hubs with strong local connections attract more external investment. Startups with predominantly female founding teams receive less funding, despite higher educational attainment. The report emphasizes the need for a more inclusive VC ecosystem.
Intel ISPC 1.30 is now available, featuring AMX (Advanced Matrix Extensions) support added to the standard library. ISPC is a variant of the C programming language designed to target Intel CPUs and GPUs.
Pluto.markets, a Danish YC-backed investment platform, has raised $6 million in a seed funding round. The round was led by Seed Capital with participation from founders of Danish unicorns such as Synthesia, Pleo, and Flatpay. The funds will be used to expand into new markets and launch new products, with the goal of disrupting the ETF market.
Kembara Fund I, managed by Mundi Ventures, has raised €750 million for investments in European deep tech startups. The fund aims for €1 billion, focusing on clean energy, AI, quantum computing, robotics, and space technologies. The EIF contributes €350 million.
The British government today launched the Advanced Nuclear Framework to attract private investment in next-generation nuclear technology for factories and datacenters. The framework aims to support the growing demand for compute power needed for artificial intelligence.
Google's selection of Innoscience for its AI platform highlights the importance of GaN (gallium nitride) manufacturing on 8-inch wafers. This technology promises to improve the efficiency and performance of artificial intelligence systems, opening new perspectives for the sector.
According to LinkedIn research, artificial intelligence is becoming standard in recruitment, shifting the focus towards hybrid skills and productivity. Recruiters are increasingly using AI to standardize hiring and find candidates faster, although difficulties persist in matching job supply and demand.
Healthtech startup Gardia has closed an €8.5 million Series A round to support the expansion of its mobile fall-detection emergency system for seniors. The funding will support expansion in the DACH region, internationalization, and strengthening B2B activities in the healthcare sector. The system is based on a wristband with automatic fall detection that works both at home and on the go, without a smartphone.
A developer created AnyTTS, a system that allows using any text-to-speech (TTS) engine with various AI chat interfaces, including ChatGPT and local LLM models. The integration happens via the clipboard, simplifying TTS usage across platforms. Currently, it only supports Windows, but the code is open for adaptations.
Siemens Digital Industries Software has announced the acquisition of Canopus AI, a move aimed at enhancing its Electronic Design Automation (EDA) stack with advanced AI-powered metrology capabilities. The integration is expected to improve semiconductor design and verification processes.
Plato, an AI-based operating system for wholesale distributors, has closed a $14.5 million seed funding round. The aim is to automate sales processes and ERP systems, addressing industry challenges such as low margins and increasing digitization. The funding will support international expansion and new features.
Valeria, a Barcelona-based startup, has raised $2 million for its payroll and workforce management platform, designed for high-turnover sectors like retail, logistics, and hospitality. The platform aims to simplify complex processes and ensure regulatory compliance, integrating with existing systems and automating onboarding via AI.
R3 Robotics (formerly Circu Li-ion) has raised €20 million to industrialize the automated dismantling of electric vehicles. The funding includes a €14 million Series A round and €6 million in European grants. The goal is to develop a platform for the automated disassembly of lithium-ion batteries, e-motors, and other components, reducing costs and improving the recovery of critical materials.
Large language models (LLMs) face complex security threats, such as sleeper-agent backdoors. These hard-to-detect attacks compromise the integrity and security of the models, opening up sci-fi-like scenarios.
Swedish startup Fintower has closed an oversubscribed €1.5 million seed round. The company develops an AI-powered SaaS platform for financial planning and analysis, aiming to modernize data-driven decision-making processes for businesses.
Finnish startup Willo has raised a €2.9 million pre-seed round to accelerate the development of its wireless power system. Unlike traditional systems, Willo's technology allows devices to be charged while moving and rotating, opening up new possibilities for robotics, industrial systems, and consumer devices. The company aims for a cable-free future, eliminating the operational constraints of manual charging.
Ghent-based HR tech company Talentguide has secured €1.3 million in funding to support the European expansion of its AI-driven skills intelligence platform. The platform helps organisations manage workforce upskilling and reskilling amid shifting labour market demands. It uses existing company data to create a skills-based job architecture and assess current employee competencies.
Berlin-based Qontext, developing an independent context layer for AI, has secured $2.7 million in pre-seed funding. The company plans to expand its platform and team to develop reusable context infrastructure, enabling AI processes to operate on reliable and continuously updated context across applications and use cases.
Microsoft introduces Paza, a project to improve automatic speech recognition (ASR) in low-resource languages. It includes PazaBench, an ASR leaderboard for 39 African languages, and Paza ASR models, optimized for six Kenyan languages. The initiative, born from Project Gecko, aims to bridge the digital and linguistic divide by developing voice technologies in collaboration with local communities and evaluating performance in real-world contexts.
The Linux 6.19 merge window introduced support for larger pages and compression with the Nouveau kernel driver, aiming to improve the performance of open-source NVIDIA drivers. Initial issues disabled this functionality, but version 7.0 should resolve them, re-enabling the use of larger pages and compression in the Mesa NVK driver.
A new study explores the use of Natural Language Processing (NLP), including Large Language Models (LLM), to automatically classify pedagogical materials against computer science curriculum guidelines. The goal is to accelerate and simplify the process of assessing content coverage.
A new study analyzes the challenges in automatically extracting medical decisions from clinical texts, revealing how linguistic variations across different decision categories negatively impact model accuracy. The analysis highlights the need for more robust extraction strategies capable of handling the stylistic diversity of medical texts.
A novel reversible deep learning model employs a conditional invertible neural network to link molecular structures and 13C NMR spectra. The network, built upon i-RevNet bijective blocks, enables spectrum prediction from structure and, conversely, the generation of structure candidates from the spectrum, addressing the one-to-many nature of spectrum-to-structure inference.
A new study analyzes the impact of differentially private training (DP-SGD) on long-tailed data, characterized by a large number of rare samples. The research highlights how DP-SGD can lead to suboptimal generalization performance, especially on these types of data, and provides a theoretical framework for understanding this phenomenon.
A new method, Iteratively Improved Program Construction (IIPC), enhances the mathematical reasoning capabilities of large language models (LLMs). IIPC iteratively refines programmatic reasoning chains, combining execution feedback with the Chain-of-thought abilities of the base model. All code is released as open source.
A new study explores the effectiveness of the Task-Method-Knowledge (TMK) framework to enhance reasoning and planning capabilities of Large Language Models (LLMs). Results show that TMK-structured prompting can significantly increase accuracy on complex tasks, bridging the gap between semantic approximation and symbolic manipulation.
Google Research has unveiled a new technique called sequential attention, aimed at making AI models leaner and faster without sacrificing accuracy. The innovation promises to reduce computational costs and improve inference efficiency.
Alphabet plans to invest US$185 billion in hardware infrastructure dedicated to artificial intelligence. The initiative aims to overcome current supply chain bottlenecks and ensure the computing capacity needed for its ambitious AI projects.
Qualcomm reported record financial results for Q1FY26. However, the company anticipates potential limitations related to memory availability in the near term, a factor that could impact deliveries and the ability to meet demand.
MediaTek projects strong growth in the cloud ASIC market, aiming for US$1 billion in revenue by 2026. The company aims to strengthen its position in this expanding sector by providing customized solutions for major cloud service providers.
A user expressed frustration with Tencent's Youtu-VL-4B model, advertised as a state-of-the-art (SOTA) solution for various computer vision tasks. Despite the promises, the released code was found to be incomplete, with key features missing and hidden in a to-do list on GitHub. The license also excludes the European Union.
According to Jensen Huang, CEO of NVIDIA, AI factories are the engine of a new wave of reindustrialization. These specialized infrastructures will be fundamental for the development and deployment of advanced AI solutions in various industrial sectors.
Skyworks Solutions signals confidence in the stability of the mobile market, while acknowledging uncertainties related to artificial intelligence and other factors. The company is preparing to navigate an evolving economic landscape, maintaining a positive outlook on its core sector.
Infineon's acquisition of Ams Osram signals a shift toward higher-value systems and AI-linked growth. The strategic move aims to strengthen Infineon's position in the market for sensors and solutions for advanced applications.
Infineon's resilience in fiscal Q1 2026 highlights how growth in the artificial intelligence sector is offsetting cyclical market pressures. The company demonstrates its ability to navigate economic challenges through a strategic focus on AI.
Qualcomm is expanding its reach beyond the smartphone market, focusing on AI-powered PCs, robotics solutions, and data center infrastructure. This strategic move aims to diversify revenue streams and capitalize on the growing demand for advanced computing capabilities across various sectors.
Semiconductor manufacturer Infineon is reallocating its production capacity towards the artificial intelligence sector, in response to a slowdown in the recovery of the automotive market. This strategic move reflects the increasing demand for high-performance chips for AI applications.
A developer has created Codag, an open-source VSCode extension that visualizes LLM workflows directly within the development environment. It supports several frameworks such as OpenAI, Anthropic, Gemini, LangChain, LangGraph, and CrewAI, along with various programming languages.
AI-Radar is an independent observatory covering AI models, local LLMs, on-premise deployments, hardware, and emerging trends. We provide daily analysis and editorial coverage for developers, engineers, and organizations exploring local AI solutions.
