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.
Elon Musk's claim of a 1 GW capacity for xAI's Colossus 2 supercomputer has been challenged. Satellite analysis suggests the site's cooling capacity indicates a significantly lower power, around 350 megawatts.
Donald Trump promised to expedite permits for nuclear-powered data centers. Jensen Huang, CEO of Nvidia, presented his vision of AI at Davos.
Higgsfield transforms simple ideas into cinematic-quality videos for social media. The platform leverages the power of advanced models like OpenAI GPT-4.1, GPT-5, and Sora 2 to automate the creation of engaging and visually stunning video content, opening new possibilities for digital creators.
PyTorch 2.10 is out today as the latest feature update to this widely-used deep learning library. The new PyTorch release continues improving support for Intel GPUs as well as for the AMD ROCm compute stack along with still driving more enhancements for NVIDIA CUDA.
Microsoft has released VibeVoice-ASR, a new model for Automatic Speech Recognition (ASR). The model is accessible via Hugging Face, opening new possibilities for developers working on voice applications. The release includes a link to the Hugging Face page and discussions on Reddit.
A new podcast analyzes ELITE, a tool Palantir is developing for ICE (Immigration and Customs Enforcement). It also discusses how AI influencers are creating fake sex tape-style photos with celebrities, and Comic-Con’s ban of AI art after artist pushback. The podcast is available on Apple Podcasts, Spotify, and YouTube.
Anthropic has introduced a new constitution for Claude, its flagship language model. This update aims to improve the model's alignment with human values and make it safer and more effective in its applications. The initiative represents a crucial step forward in the responsible development of artificial intelligence.
Intel has announced the end-of-life (EOL) for its 12th Generation Alder Lake and 4th Generation Xeon Sapphire Rapids processors. Customers will have a limited time to place final orders for these hybrid CPUs, marking a significant shift in Intel's product roadmap.
OpenAI is trying to alleviate concerns about its new Stargate datacenters. The company promises plans that take into account local needs, minimizing the environmental impact and the impact on electricity costs. The initiative comes at a time of increasing attention to the energy consumption linked to artificial intelligence.
A new model named GLM-OCR from Z.ai has been spotted on GitHub. The finding was reported on Reddit, in the LocalLLaMA subreddit, via a post including an image and links to the discussion and the original resource. Further details on the model's capabilities or technical specifications are currently unavailable.
XDG-Desktop-Portal 1.21 is now available for testing with the latest features for this portal frontend service to Flatpak. Key updates include support for Linyaps applications and a reduced motion setting, aimed at improving user experience and accessibility.
Nvidia CEO Jensen Huang confirmed that his company has overtaken Apple as TSMC's biggest customer, becoming its top client after more than 20 years. This shift underscores Nvidia's growing prominence in the semiconductor industry.
YouTube is introducing a feature that will allow content creators to make Shorts using AI versions of themselves. Viewers might soon see AI avatars of their favorite YouTubers while scrolling through Shorts feeds.
The NVIDIA GB10 superchip, designed for AI, has been tested in traditional Linux scenarios to evaluate its CPU performance. Phoronix benchmarks compare the GB10 with the AMD Ryzen AI Max+ "Strix Halo" within the Framework Desktop, offering a glimpse into the chip's capabilities in workloads other than AI.
Palantir CEO Alex Karp has voiced a potentially controversial opinion on the impact of artificial intelligence (AI) on immigration. According to Karp, AI could reduce the need for immigration in Western economies. His claims have sparked heated debate, given the sensitivity of the topic and the resulting economic and social implications.
A bug in GLM-4.7-Flash-GGUF causing looping and poor outputs has been fixed. Users are advised to redownload the model for significantly improved results. Z.ai has suggested optimal parameters for various use cases, including general use and tool-calling. The update is available on Hugging Face.
OpenAI is on track to announce its first hardware device, possibly earbuds, in 2026. OpenAI Chief Global Affairs Officer Chris Lehane said that the company plans to unveil its first hardware in the second half of this year. This move marks a significant expansion for OpenAI, traditionally focused on developing language models and software.
A new study by Vanguard forecasts that artificial intelligence (AI) will significantly impact productivity, comparable to the personal computer. AI will augment human capabilities rather than completely replace them, leading to a transformation of work and economic growth. Service sectors, such as healthcare and education, will see the greatest benefits from AI-driven automation.
We compared the AI models from Google (Gemini 3.2 Fast) and OpenAI (ChatGPT 5.2) to evaluate their performance. The tests, based on complex prompts, aim to simulate the standard user experience, that is, those who do not pay for subscriptions. The analysis combines objective evaluations and subjective impressions, updating the comparative tests carried out in 2023.
Here's how to get GLM 4.7 working on llama.cpp using Flash Attention for improved performance. The guide includes configuration details and a link to a specific Git branch. Note that quantizations may need to be recreated to avoid nonsensical outputs.
Nvidia CEO Jensen Huang is heading to China in late January for a customary Lunar New Year visit. The trip gains importance as it coincides with negotiations regarding the quantity of H200 GPUs Beijing will be allowed to import, amid U.S. export restrictions on advanced chips to China.
Adobe is integrating artificial intelligence tools into Acrobat, offering new features such as automatic podcast summary generation, presentation creation, and file editing via text prompts. The goal is to simplify and speed up user workflows.
OpenAI aims to reassure investors ahead of its potential initial public offering (IPO), demonstrating a clear correlation between computing power and revenue growth. The company continues to invest heavily in infrastructure, with expenditure currently outweighing income, particularly for the construction of a 10GW infrastructure.
Microsoft CEO Satya Nadella warns that artificial intelligence must generate benefits for a broad segment of the population, otherwise it risks losing social permission and turning into a speculative bubble. A wider impact is needed to prevent the benefits from being concentrated in the hands of a few.
Zanskar has raised $115 million to find about a dozen geothermal resources throughout the U.S. West. The goal is to power the grid with clean energy, exploiting previously unexplored potential. The initiative is expected to significantly contribute to the energy transition.
Large language models (LLMs) continue to be vulnerable to prompt injection attacks, a technique that tricks AI into performing unauthorized actions. The difficulty lies in their inability to understand context as a human would, making them susceptible to manipulations that bypass security measures. New approaches are needed to effectively protect these systems.
Preply, a Ukrainian-founded language learning marketplace, has raised $150 million in Series D funding led by WestCap, valuing the company at $1.2 billion. Preply connects over 100,000 tutors with learners in 180 countries, offering one-on-one lessons in more than 90 languages. The new capital will be used to advance AI capabilities and expand the engineering team.
A fix for an issue related to GLM 4.7 Flash has been merged into llama.cpp. In parallel, FA (Fused Attention) support for CUDA is under development, aiming to further improve performance and efficiency in using NVIDIA GPUs for language model inference.
File Brain is an open-source search engine that indexes local files and allows searching using natural language. It supports multilingual semantic search, built-in OCR, and is available for Windows and Linux. The goal is to overcome the limitations of keyword-based search tools by understanding document content.
In 2025, consumer spending on mobile applications surpassed that of mobile games. The adoption of AI-powered apps was the primary driver of this growth, marking a significant shift in digital spending habits.
Language learning marketplace Preply is now valued at $1.2 billion after raising $150 million. This milestone marks a new chapter for the 14-year-old company and embodies the resilience of the Ukrainian tech sector, where Preply has its roots.
Microsoft CEO Satya Nadella says datacenter location is "the least important thing" for AI sovereignty. Ownership of models and embedded corporate knowledge matters more than server location, according to Nadella.
OpenAI is committed to ensuring that electricity prices do not increase in the communities where it builds its Stargate data centers. The company will fund grid upgrades and flexible load management systems to reduce stress on the energy supply. The goal is to ensure that the expansion of AI infrastructure does not burden consumers.
Silicio Valley’s battle against AI regulation is already shaping the next US election cycle. Pro-AI Super PACs are preparing to invest heavily in the midterm elections.
An Amazon customer was scammed: instead of an RTX 5080 graphics card, they received a relabelled RTX 5060 Ti. The package was sold and shipped by Amazon, suggesting a possible return switcheroo. The deception was spotted due to the 8-pin power connector, typical of lower-end cards, instead of the 16-pin connector of the RTX 5080.
A Linux kernel patch aims to significantly reduce wake-up latency on modern Intel Xeon servers. The modification, involving a single line of code, aims to optimize performance in scenarios where responsiveness is critical, especially with NOHZ_FULL configurations and the Linux menu governor.
This comparison matrix presents decision axes, trade-offs, and constraints solely for evaluation purposes. It does not constitute a recommendation, endorsement, or ranking of deployment models. Final decisions should be guided by your organization's specific requirements, regulatory context, and risk tolerance.
AI cost efficiency clashes with data sovereignty, forcing companies to rethink their risk frameworks. The case of DeepSeek, a Chinese AI lab, raises concerns about data sharing with state intelligence services. This requires stricter governance, especially in sectors like finance and healthcare, where transparency on data provenance is crucial to avoid violations and reputational damage.
OpenAI has set 2026 as the key year for the widespread adoption of truly usable artificial intelligence solutions. The company is also looking at entering the hardware market and diversifying its revenue streams, in a context of increasing competition and demand for increasingly advanced computational resources to support the development of its models.
Citi has undertaken an internal initiative to integrate artificial intelligence into the daily work of its employees. Approximately 4,000 people, from various business sectors, have been trained to use approved AI tools. The goal is to improve efficiency and distribute AI knowledge throughout the organization, reducing reliance on specialist teams and promoting responsible use of technology.
Berlin-based climate fintech Cloover has secured over $1.2 billion in funding to accelerate the rollout of its software and financing platform across Europe. The financing package includes Series A equity and a substantial debt facility. The company aims to become the “Shopify of Energy”.
Venture capital firm Sequoia Capital is reportedly considering an investment in Anthropic, an artificial intelligence company. This move could signal a shift in the fund's investment strategy, with a greater focus on emerging technologies in the AI field. Anthropic is known for developing advanced language models and innovative solutions in the sector.
Unimicron and Nan Ya anticipate increased profits for December 2025, driven by strong demand for PCBs (Printed Circuit Boards) fueled by the artificial intelligence sector. This increase underscores the crucial role of PCB manufacturers in the rapidly expanding AI ecosystem.
Fortune Electric is experiencing tailwinds from lower US tariffs and rising demand for AI data centers. These combined factors create a favorable environment for the company, improving growth and profitability prospects in the rapidly expanding AI sector.
Taiwan's exports reached record highs in December, driven by strong demand for AI servers and chips. This highlights the island's key role in the global semiconductor industry and its growing importance in the AI sector.
OpenAI introduces "Edu for Countries", a new initiative designed to support governments in adopting artificial intelligence. The goal is to modernize education systems and prepare the workforce of the future, providing tools and resources to integrate AI into learning and professional development.
A new report highlights the stark differences in the adoption of advanced artificial intelligence across countries. New strategic initiatives are outlined to help nations maximize the productivity gains from AI. The goal is to bridge the existing gap and foster more uniform economic growth through the effective implementation of these technologies.
An article explores how corporate knowledge, if poorly structured and rigidly transferred, can transform from an asset into a disadvantage, both for companies and employees. The onboarding process is crucial: inadequate information management can compromise the effectiveness of new hires and, over time, erode the value of the company's human capital. Well-organized corporate knowledge is essential for long-term success.
Aerospace company SWISSto12 has secured €73 million in financial support from ESA to accelerate the development of the HummingSat platform. The funding will be used to industrialize HummingSat, increase manufacturing capacity, and promote new product innovations. This initiative addresses the increasing global demand for cost-effective, agile, and sovereign communications in both government and commercial sectors. The first launch is scheduled for 2027.
Berlin-based Cloover has secured $1.2 billion in debt and equity funding to develop an AI-powered operating system aimed at accelerating the energy transition. The funds will support expansion into new European markets and further development of its platform with AI-driven automation and financing capabilities, with the ambition of connecting manufacturers, installers, and investors, promoting distributed energy projects.
Fracttal, a Madrid-based company specializing in AI-powered maintenance solutions, has closed a $35 million funding round led by Riverwood Capital. The investment will support the company's continued growth, product development, and global expansion. The Fracttal One platform centralizes maintenance operations through integrations with enterprise systems, third-party IoT sensors, and proprietary IoT hardware, improving efficiency, safety, and sustainability.
Antidote, a provider of AI-based billing compliance software for law firms, has raised $5 million in a seed funding round. The funding will support the advancement of its platform and expand its presence in the US, aiming to reduce billing errors and automate processes.
The Davos 2026 Forum will feature artificial intelligence as a key topic. Global leaders will discuss crucial issues such as the necessary computing power, the control of algorithms, and the ethical and social implications arising from its development. The event promises to be a turning point in defining the future of AI and its impact on the world.
Two recent rocket launch failures in China, highlighted by Galactic Energy, underscore the risks and challenges inherent in the country's growing space ambitions. These incidents raise questions about the reliability of Chinese launch technologies and could have repercussions for future space exploration plans and satellite deliveries. Reliability remains a critical factor in the aerospace sector.
Taiwan benefits from favorable trade tariffs, but faces a decrease in direct investments in China and Vietnam. The geopolitical situation and regional economic dynamics influence the investment strategies of Taiwanese companies, which are seeking new growth opportunities.
The electric vehicle market is showing signs of division, while competition in the edge AI sector is intensifying. New analysis reveals emerging trends and the challenges companies face to succeed in these rapidly evolving sectors. Insights into winning strategies and key factors for growth.
Fortune Electric and Nan Ya are aiming to expand their presence in the US transformer market, capitalizing on the urgent need to upgrade the country's aging grid infrastructure. Demand, driven by large-scale AI computing, is outpacing domestic production capacity, creating opportunities for Taiwanese suppliers.
An enthusiast shares progress on building a language model (LM) from scratch. After stabilizing the system, the focus shifted to training, revealing the need for a significantly higher number of steps to achieve optimal results. Despite initial challenges related to using DataParallel on Windows, the model shows promising language generation capabilities, with a nearly perfect sentence structure.
According to Digitimes sources, the easing of US tariffs is slowing down the shift of supply chains towards Southeast Asia. This unexpected change is significantly impacting the strategies of companies that aimed to diversify their production to avoid global trade tensions.
The maintainer of the popular open-source data transfer tool Curl has ended the project’s bug bounty program, following a surge of AI-generated submissions. The initiative had become unmanageable due to the difficulty of assessing automated contributions. The maintainer hopes hackers will still send bug reports and promises to continue shaming the "silly ones."
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.
