Look across the history of consumer tech and a pattern appears. Ownership gives way to services, and services become subscriptions. We went from stacks of DVDs to streaming movies online, from external drives for storing data and backups to cloud drives, from MP3s on a player to Spotify subscriptions, from one time software licenses to recurring plans. But when AI arrived, it skipped the ownership phase entirely. Intelligence came as a service, priced per month or per million tokens. No ownership, no privacy. Just a $20 a month fee.

A device like Olares One rearranges that relationship. It compresses a full AI stack into a desktop sized box that behaves less like a website and more like a personal studio. You install models the way you once installed apps. You shape its behavior over time, training it on your documents, your archives, your creative habits. The result is an assistant that feels less rented and more grown, with privacy, latency, and long term cost all tilting back toward the owner.

Designer: Olares

Click Here to Buy Now: $2,899 $3,999 (28% off) Hurry! Only 15/320 units left!

The pitch is straightforward. Take the guts of a $4,000 gaming laptop, strip out the screen and keyboard, put everything in a minimalist chassis that looks like Apple designed a chonky Mac mini, and tune it for sustained performance instead of portability. Dimensions are 320 x 197 x 55mm, weighs 2.15 kg without the PSU, and the whole package pulls 330 watts under full load. Inside sits an Intel Core Ultra 9 275HX with 24 cores running up to 5.4 GHz and 36 MB of cache, the same chip you would find in flagship creator laptops this year. The GPU is an NVIDIA GeForce RTX 5090 Mobile with 24 GB of GDDR7 VRAM, 1824 AI TOPS of tensor performance, and a 175W max TGP. Pair that with 96 GB of DDR5 RAM at 5600 MHz and a PCIe 4.0 NVMe SSD, and you have workstation level compute in a box smaller than most soundbars.

Olares OS runs on top of all that hardware, and it is open source, which means you can audit the code, fork it, or wipe it entirely if you want. Out of the box it behaves like a personal cloud with an app store containing over 200 applications ready to deploy with one click. Think Docker and Kubernetes, but without needing to touch a terminal unless you want to. The interface looks clean, almost suspiciously clean, like someone finally asked what would happen if you gave a NAS the polish of an iPhone. You get a unified account system so all your apps share a single login, configurable multi factor authentication, enterprise grade sandboxing for third party apps, and Tailscale integration that lets you access your Olares box securely from anywhere in the world. Your data stays on your hardware, full stop.

I have been tinkering with local LLMs for the past year, and the setup has always been the worst part. You spend hours wrestling with CUDA drivers, Python environments, and obscure GitHub repos just to get a model running, and then you realize you need a different frontend for image generation and another tool for managing multiple models and suddenly you have seven terminal windows open and nothing talks to each other. Olares solves that friction by bundling everything into a coherent ecosystem. Chat agents like Open WebUI and Lobe Chat, general agents like Suna and OWL, AI search with Perplexica and SearXNG, coding assistants like Void, design agents like Denpot, deep research tools like DeerFlow, task automation with n8n and Dify. Local LLMs include Ollama, vLLM, and SGIL. You also get observability tools like Grafana, Prometheus, and Langfuse so you can actually monitor what your models are doing. The philosophy is simple. String together workflows that feel as fluid as using a cloud service, except everything runs on metal you control.

Gaming on this thing is a legitimate use case, which feels almost incidental given the AI focus but makes total sense once you look at the hardware. That RTX 5090 Mobile with 24 GB of VRAM and 175 watts of power can handle AAA titles at high settings, and because the machine is designed as a desktop box, you can hook it up to any monitor or TV you want. Olares positions this as a way to turn your Steam library into a personal cloud gaming service. You install your games on the Olares One, then stream them to your phone, tablet, or laptop from anywhere. It is like running your own GeForce Now or Xbox Cloud Gaming, except you own the server and there are no monthly fees eating into your budget. The 2 TB of NVMe storage gives you room for a decent library, and if you need more, the system uses standard M.2 drives, so upgrades are straightforward.

Cooling is borrowed from high end laptops, with a 2.8mm vapor chamber and a 176 layer copper fin array handling heat dissipation across a massive 310,000 square millimeter surface. Two custom 54 blade fans keep everything moving, and the acoustic tuning is genuinely impressive. At idle, the system sits at 19 dB, which is whisper quiet. Under full GPU and CPU load, it climbs to 38.8 dB, quieter than most gaming desktops and even some laptops. Thermal control keeps things stable at 43.8 degrees Celsius under sustained loads, which means you can run inference on a 70B model or render a Blender scene without the fans turning into jet engines. I have used plenty of small form factor PCs that sound like they are preparing for liftoff the moment you ask them to do anything demanding, so this is a welcome change.

RAGFlow and AnythingLLM handle retrieval augmented generation, which lets you feed your own documents, notes, and files into your AI models so they can answer questions about your specific data. Wise and Files manage your media and documents, all searchable and indexed locally. There is a digital secret garden feature that keeps an AI powered local first reader for articles and research, with third party integration so you can pull in content from RSS feeds or save articles for later. The configuration hub lets you manage storage, backups, network settings, and app deployments without touching config files, and there is a full Kubernetes console if you want to go deep. The no CLI Kubernetes interface is a big deal for people who want the power of container orchestration but do not want to memorize kubectl commands. You get centralized control, performance monitoring at a glance, and the ability to spin up or tear down services in seconds.

Olares makes a blunt economic argument. If you are using Midjourney, Runway, ChatGPT Pro, and Manus for creative work, you are probably spending around $6,456 per year per user. For a five person team, that balloons to $32,280 annually. Olares One costs $2,899 for the hardware (early-bird pricing), which breaks down to about $22.20 per month per user over three years if you split it across a five person team. Your data stays private, stored locally on your own hardware instead of floating through someone else’s data center. You get a unified hub of over 200 apps with one click installs, so there are no fragmented tools or inconsistent experiences. Performance is fast and reliable, even when you are offline, because everything runs on device. You own the infrastructure, which means unconditional and sovereign control over your tools and data. The rented AI stack leaves you as a tenant with conditional and revocable access.

Ports include Thunderbolt 5, RJ45 Ethernet at 2.5 Gbps, USB A, and HDMI 2.1, plus Wi-Fi 7 and Bluetooth 5.4 for wireless connectivity. The industrial design leans heavily into the golden ratio aesthetic, with smooth curves and a matte aluminum finish that would not look out of place next to a high end monitor or a piece of studio equipment. It feels like someone took the guts of a $4,000 gaming laptop, stripped out the compromises of portability, and optimized everything for sustained performance and quietness. The result is a machine that can handle creative work, AI experimentation, gaming, and personal cloud duties without breaking a sweat or your eardrums.

Olares One is available now on Kickstarter, with units expected to ship early next year. The base configuration with the RTX 5090 Mobile, Intel Core Ultra 9 275HX, 96 GB RAM, and 2 TB SSD is priced at a discounted $2,899 for early-bird backers (MSRP $3,999). That still is a substantial upfront cost, but when you compare it to the ongoing expense of cloud AI subscriptions and the privacy compromises that come with them, the math starts to make sense. You pay once, and the machine is yours. No throttling, no price hikes, no terms of service updates that quietly change what the company can do with your data. If you have been looking for a way to bring AI home without sacrificing capability or convenience, this is probably the most polished attempt at that idea so far.

Click Here to Buy Now: $2,899 $3,999 (28% off) Hurry! Only 15/320 units left!

The post This $2,899 Desktop AI Computer With RTX 5090M Lets You Cancel Every AI Subscription Forever first appeared on Yanko Design.

We have come a long way from the computers the size of entire rooms to the sleek personal computers that sit comfortably on our desks. The evolution of computing has consistently pushed toward smaller form factors and greater efficiency. The Mac mini, for example, illustrates how compact modern PCs have become. Yet the question persists: how miniature can a powerful computing device truly be? A recent Guinness World Records certification offers a striking answer.

Tiiny AI, a US-based deep-tech startup, has unveiled the Pocket Lab, officially verified as the “world’s smallest personal AI supercomputer.” This palm-sized device, no larger than a typical power bank, is capable of running large language models (LLMs) with up to 120 billion parameters entirely on-device, without relying on cloud servers or external GPUs.

Designer: Tiiny AI

At its core, the Pocket Lab aims to make advanced artificial intelligence both personal and private. Traditional AI systems often depend on cloud infrastructure, which can raise concerns around data privacy, latency, and carbon emissions associated with large server farms. The Pocket Lab addresses these issues by enabling fully offline AI computation. All processing, data storage, and inference happen locally on the device, reducing dependence on internet connectivity or cloud resources.

Despite its compact size, measuring 14.2 × 8 × 2.53 centimeters and weighing roughly 300 grams, this mini supercomputer delivers noteworthy computing power. The system operates within a typical 65-watt energy envelope, comparable to a conventional desktop PC, yet manages to support extensive AI workloads. The hardware architecture combines a 12-core ARMv9.2 CPU with a custom heterogeneous module that includes a dedicated Neural Processing Unit (NPU), together achieving approximately 190 TOPS (tera-operations per second) of AI compute performance. This configuration is backed by 80 GB of LPDDR5X memory and a 1 TB solid-state drive, allowing large AI models to run efficiently without external accelerators.

Two key technologies underpin the Pocket Lab’s ability to run large models so efficiently in such a small package. TurboSparse improves inference efficiency through neuron-level sparse activation, reducing computational overhead while preserving model intelligence. PowerInfer, an open-source heterogeneous inference engine with a significant developer following, dynamically distributes workloads across the CPU and NPU, delivering server-grade performance at far lower power and cost than traditional GPU-based solutions.

In practical terms, the Pocket Lab supports a wide ecosystem of open-source AI models and tools. Users can deploy popular LLMs such as GPT-OSS, Llama, Qwen, DeepSeek, Mistral, and Phi, alongside agent frameworks and automation tools, all with one-click installation. This broad software compatibility extends the device’s usefulness beyond enthusiasts and hobbyists to developers, researchers, professionals, and students.

By storing all user data and interactions locally with bank-level encryption, the Pocket Lab also emphasizes privacy and long-term personal memory. This feature contrasts with many cloud-based AI services that retain data on remote servers. Tiiny AI plans to showcase the Pocket Lab at CES 2026, but has not yet disclosed full details on pricing or release dates.

The post Record setting Pocket Lab shrinks a full AI supercomputer into the size of a power bank first appeared on Yanko Design.

Vous avez, j’imagine, probablement des dizaines de milliers de photos sur votre disque dur. Ça représente peut-être quelques centaines de Go, peut-être 1 To si vous êtes à l’aise en espace de stockage. C’est beaucoup ?

Pas tant que ça si on pense un peu à votre cerveau. Lui, il stocke depuis toujours des décennies de souvenirs dans environ 1,5 kg de matière organique qui consomme moins qu’une ampoule LED.

Comment est-ce qu’il fait ?

Hé bien, une équipe du Weizmann Institute of Science vient peut-être de le découvrir et au passage, changer la compression d’images telle qu’on la connaît.

Le projet s’appelle Brain-IT , et leur idée c’est de reconstruire des images à partir des signaux fMRI (imagerie par résonance magnétique fonctionnelle) de votre cerveau. En gros, ils scannent votre activité cérébrale pendant que vous regardez une image, et ils arrivent à reconstruire ce que vous avez vu. Le papier scientifique est dispo sur arXiv si vous parlez leur langue.

Évidemment, ce genre de recherche, c’est pas nouveau mais Brain-IT est plutôt un franc succès car le process permet d’obtenir les mêmes résultats que les méthodes précédentes avec seulement 1 heure de données fMRI, contre 40 heures pour les autres approches.

En gros, ça représente 97,5% de données en moins pour obtenir le même résultat. Trop fort non ?

En fait, si Brain-IT peut faire ça, c’est parce que les scientifiques ont découvert comment votre cerveau compresse les images de manière hyper efficace. Et d’ailleurs, ce truc pourrait bien inspirer de nouveaux algorithmes de compression pour nos ordis.

Brain-IT utilise en fait ce qu’ils appellent un “Brain Interaction Transformer” (BIT). C’est un système qui identifie des “clusters fonctionnels” de voxels cérébraux. Un voxel, c’est l’équivalent d’un pixel mais en 3D, et chaque voxel représente environ 1 million de cellules dans votre cerveau.

Le truc génial, c’est que ces clusters fonctionnels sont partagés entre différentes personnes, comme si nous avions tous la même bibliothèque de “primitives visuelles” câblée dans nos têtes. Ce sont des schémas de base que notre cerveau utilise pour reconstruire n’importe quelle image.

Brain-IT reconstruit donc les images en deux passes. D’abord les structures de bas niveau (les formes, les contours), puis les détails sémantiques de haut niveau (c’est un chat, c’est un arbre, c’est votre tante Huguette). C’est un peu comme le JPEG progressif que l’on voyait s’afficher lentement avec nos modem 56K, mais en infiniment plus smart.

Du coup, si on comprend comment le cerveau compresse les images, on pourrait créer de nouveaux formats vidéo ultra-légers. Imaginez un Netflix ou un Youtube qui streame en “brain-codec” à 1/40e de la bande passante actuelle. Ça changerait pas mal de choses… Et c’est pareil pour l’IA générative car actuellement, on entraîne des modèles avec des millions d’images durant des jours alors que notre cerveau, lui, apprend à reconnaître un visage en quelques expositions.

Et grâce à ses modèles de diffusion, Brain-IT est même capable de reconstruire visuellement ce que voit le cerveau ! Par contre, Brain-IT n’a pour le moment été testé que sur des images “vues” et pas des choses imaginées…

Mais les scientifiques n’écartent pas l’idée que ce soit possible donc ce n’est qu’une question de temps avant qu’on puisse capturer en image ses rêves par exemple.

Voilà, j’ai trouvé ça cool parce que ça montre que notre cerveau est probablement le meilleur système de compression d’images jamais créé et qu’on commence à peine à comprendre comment il fonctionne.

Merci Dame Nature !

Voilà, si vous voulez creuser, le code et les détails techniques sont sur la page du projet Brain-IT , et le paper complet est dispo sur arXiv .

Source

Vous savez ce qui est encore plus classe qu’un hacker à capuche dans sa cave ? Et bien c’est certainement quand un groupe de hackers allemands décide dans les années 80 de faire trembler les gouvernements, de défier le KGB et d’inventer au passage la moitié des techniques de cybersécurité qu’on utilise encore aujourd’hui.

Bienvenue dans l’univers du Chaos Computer Club, une organisation incroyable qui a façonné notre monde numérique moderne.

NVIDIA launched its most affordable RTX 50-series GPU yet for laptops, and a desktop version of the RTX 5050 is also on the way, starting at $249.

AMD has listed a Ryzen 5 9600X3D CPU in official documentation for the first time, suggesting that it could have a new budget champion headed our way in 2025.

Lenovo's latest flagship desktop all-in-one is the first Copilot+ all-in-one PC, and it trumps the competition, showing how dire the AiO Windows PC space is.

There's a new black screen bug affecting RTX 5060 and RTX 5060 Ti GPUs, but NVIDIA has taken the initiative by explaining and, more importantly, offering a fix.

Qualcomm's next-gen Snapdragon X chips are on the way. However, it doesn't look like we'll get an unveil until October 2025, pointing to a 2026 launch.

AMD's Ryzen 9 7900X is one of Team Red's most powerful CPUs for gaming and work, and it's now down to $325 today only. A 1TB SSD is included for free.

The Beelink EQI12 is a fantastic little computer that can handle your work, school, or home workloads with ease. Right now, it's very affordable while this Memorial Day deal lasts.

A new leak points to an RTX 5060 launch on May 19, but the leak also points to the review embargo lifting the same day. Does that raise alarms for anyone else?

AMD could be prepping an RX 9060 XT announcement at Computex, with launch to follow in early June. Here's what you need to know about leaks and rumors so far.

NVIDIA's drivers have been shaky since the launch of RTX 50-series GPUs, but the latest update has a long list of fixes. Unfortunately, they don't work for everyone.

New GeForce RTX 5070 Ti GPUs sold out in seconds and are again suffering a scalping issue with inflated prices above MSRP.

NVIDIA's RTX 5070 Ti unsurprisingly sold out in seconds, but that doesn't mean you can't get the new GPU on launch day. I found these pre-built PCs with the GPU.

NVIDIA's RTX 5070 Ti launches today, but it's not expected to be a smooth process. I found the best places to shop, with a look at whether it's worth it.

After weeks of delay, AMD is ready to reveal its RDNA 4 graphics cards on February 28. Will it be a success? Here are 5 reasons why it might fall flat.

Arm has withdrawn its threat to end a licensing agreement with Qualcomm. The news comes just weeks after a major court ruling on a case between the two tech giants.