AI-RADAR | Monitoraggio Intelligenza Artificiale

Dec 29 2025

LLM

New Generation of Multilingual Models: Gamayun Challenges Giants

Gamayun is a new multilingual LLM model that has gained attention for its innovative pre-training strategy, achieving impressive results despite having a smaller training budget than competitors.

→

Dec 29 2025

LLM

The Unified Definition of Illusion: A Global Model, Stupid

A unified theory of illusion has been developed, providing a common framework for evaluating model performance and distinguishing between different types of errors.

→

Dec 29 2025

LLM

dUltra: Ultra-Fast Diffusion Language Models via Reinforcement Learning

Meta has announced the creation of dUltra, a new learning framework to improve diffusion model performance. This new approach uses reinforcement learning to optimize parallel decoding.

→

Dec 29 2025

LLM

Physics-Informed Neural Solvers for Periodic Quantum Eigenproblems

→

Dec 29 2025

LLM

Feasible strategies in three-way conflict analysis with three-valued ratings

A new study explores how large language models can help understand and resolve complex conflicts. Researchers have developed a new approach to analyze disputes and identify feasible strategies.

→

Dec 29 2025

LLM

Three-way conflict analysis based on alliance and conflict functions

A new method for conflict analysis has been developed by separating the alliance and conflict functions. This approach can help to better understand the relationships between agents and issues at stake.

→

Dec 28 2025

Hardware

World’s smallest autonomous robots are 'smaller than a grain of salt,' cost one penny apiece — researchers expect new micron-scale fully-programmable robots to be used in medicine, microscale manufacturing, and other areas

Research teams from the University of Pennsylvania and the University of Michigan have developed fully programmable, autonomous robots 'smaller than a grain of salt'. Researchers expect these new micron-scale robots to be used in medicine, microscale manufacturing, and other areas

→

Dec 28 2025

LLM

OpenAI seeks new Head of Preparedness

→

Dec 28 2025

Altro

Rainbow Six Siege under siege by hackers, Ubisoft forced to take all servers offline

Players of Rainbow Six Siege have been targeted by hackers who have made the game's online services unusable. Players have received billions of credits and exclusive items, while Ubisoft has had to take drastic measures to protect its community.

→

Dec 28 2025

Hardware

Zephyr GPU manufacturer confirms dead RDNA 2 GPU chips

→

Dec 28 2025

Hardware

Modders Challenge: RTX 5080 with 32GB VRAM, but What Does it Mean for Gamers?

The modding community is exploring new possibilities with the Nvidia RTX 5080. While this GPU may improve specific performance, its implications for gamers seeking a smooth gaming experience are unclear.

→

Dec 27 2025

LLM

ChatGPT: OpenAI uses data to promote sponsored content

OpenAI is integrating sponsored content into ChatGPT's responses, a new advertising strategy within the platform.

→

Dec 27 2025

Hardware

Some Japanese shops start rationing GPUs

→

Dec 27 2025

Hardware

Nvidia GeForce RTX 5090 reportedly caught fire despite using original cable

User reports Nvidia GeForce RTX 5090 power connector reportedly caught fire despite using original cable

→

Dec 26 2025

LLM

Artificial Intelligence Models Will Dominate the Market in 2026

The world of technology is expected to see significant changes in the coming year, with artificial intelligence models playing a key role.

→

Dec 26 2025

Market

Asus denies memory manufacturing rumors

Asus denies entering the memory market within six months

→

Dec 26 2025

LLM

China's control of battery supply chains is becoming a critical risk for U.S. military power and AI initiatives

La Cina controlla gran parte delle catene di approvvigionamento delle batterie, una preoccupazione che sta diventando sempre più critica per le forze armate statunitensi e le iniziative di intelligenza artificiale.

→

Dec 26 2025

Hardware

Intel Revolutionizes Processor Production with Gigantic Multi-Chiplet Packages

Intel has introduced a new type of multi-chiplet package that surpasses the size of the largest AI processors by 12 times, utilizing advanced technologies like HBM5 and SRAM 14A/18A.

→

Dec 26 2025

LLM

Elon Musk says xAI will have more AI compute than everyone else combined within five years

Elon Musk claims that xAI will have more computing power than everyone else combined within five years

→

Dec 26 2025

Hardware

Advanced robotics: how lobster exoskeletons turn into robots

Swiss scientists have demonstrated the use of lobster tails as robotic appendages. This innovation combines food waste with synthetic components, bringing a new frontier to sustainable robotics.

→

Dec 26 2025

Hardware

Amazon Christmas miracle: two orders of Samsung 9100 Pro SSDs turn into 20 SSDs worth over $5,100

A user received two boxes of expensive Samsung 9100 Pro SSDs, worth over $5,100, due to Amazon's packaging error. The situation has been described as a 'Christmas miracle' amidst the ongoing component crisis.

→

Dec 26 2025

LLM

New Turn in AI-Assisted Healthcare Models

Meta has announced the launch of its new AI-assisted healthcare model, Erkang-Diagnosis-1.1. The model combines a hybrid approach with pre-training and return generation to create a secure, reliable, and professional AI health advisor.

→

Dec 26 2025

LLM

MicroProbe: Efficient Reliability Assessment for Foundation Models with Minimal Data

Researchers have developed MicroProbe, a new technology that enables reliability assessment of foundation models using only minimal data.

→

Dec 26 2025

LLM

New Technology for Better Language Model Understanding

Researchers have developed a new technology that enables language models to better understand context and relationships between concepts. This innovation could revolutionize the approach to text comprehension problems.

→

Dec 26 2025

Frameworks

Quantum-Inspired Multi Agent Reinforcement Learning for Exploration Exploitation Optimization in UAV-Assisted 6G Network Deployment

Researchers have developed a new framework that combines AI with quantum physics to optimize 6G network management. The approach, called QI MARL, promises significant improvements in scalability and efficiency.

→

Dec 26 2025

LLM

Artificial Intelligence for Smart Home Lighting Optimization

Artificial Intelligence is revolutionizing smart home lighting optimization. The new BitRL-Light model combines Llama with the Deep Q-Network to optimize energy consumption and improve user comfort.

→

Dec 25 2025

LLM

Google allows users to change their Gmail address

Google allows users to change their Gmail address without creating a new account

→

Dec 25 2025

Hardware

Retired Navy nuclear reactors may soon power AI data centers

A Texas energy startup proposes repurposing retired U.S. Navy nuclear reactors for use in AI data centers

→

Dec 25 2025

Hardware

China chipmaker SMIC raises wafer prices by about 10% as memory demand tightens capacity

Chinese chipmaker SMIC raises wafer prices by about 10% as memory demand tightens capacity

→

Dec 25 2025

Hardware

Corsair RM850x SHIFT: Platinum performance at Gold pricing

The Corsair RM850x SHIFT is an innovative power supply that relocates modular connectors to the chassis side for superior cable management. It delivers Cybenetics Platinum efficiency under a Gold-level marketing badge, making it suitable for builders willing to verify case compatibility.

→

Dec 25 2025

Hardware

Qualcomm Snapdragon X Elite's Linux benchmarks show regressions, like Intel Tiger Lake chips five years ago

Latest Linux benchmarks on Qualcomm Snapdragon X Elite show significant regressions, similar to Intel Tiger Lake chips five years ago

→

Dec 25 2025

Altro

The nostalgia of floppy disks is back: a pre-paid cash card with a floppy disk shape

A Taiwanese company has released a new pre-paid cash card that resembles a floppy disk, sparking nostalgia for the old technology.

→

Dec 25 2025

LLM

Uncovering Competency Gaps in Large Language Models and Their Benchmarks

La valutazione dei grandi modelli linguistici (LLM) si basa pesantemente su benchmarks standardizzati. Questi benchmarks offrono metriche aggregate utili per una data capacità, ma queste metriche aggregate possono nascondere (i) aree particolari dove i modelli sono deboli ('lacune del modello') e (ii) distorsioni nella copertura dei benchmark stessi ('lacune del benchmark'). Presentiamo un nuovo metodo che utilizza autoencoditori sparsi (SAEs) per scoprire automaticamente entrambi tipi di lacuna. Sfruttando le attivazioni concettuali degli SAE e calcolando i punteggi dei prestazioni salienza-weighted in base a dati benchmark, il metodo pone l'evaluzione sulle rappresentazioni interne del modello ed permette una comparazione tra i benchmarks.

→

Dec 25 2025

LLM

Advanced Language Models for Enhancing Lung Cancer Treatment Outcome Prediction

Predicting treatment outcomes for lung cancer remains a challenge due to the sparsity, heterogeneity, and information overload of real-world electronic health data.

→

Dec 25 2025

LLM

Learning Evolving Latent Strategies for Multi-Agent Language Systems without Model Fine-Tuning

This study proposes a multi-agent language framework that enables continual strategy evolution without fine-tuning the language model's parameters. The core idea is to liberate the latent vectors of abstract concepts from traditional static semantic representations, allowing them to be continuously updated through environmental interaction and reinforcement feedback.

→

Dec 25 2025

LLM

Zero-Training Temporal Drift Detection for Transformer Sentiment Models: A Comprehensive Analysis on Authentic Social Media Streams

A recent study analyzes the stability of transformer-based sentiment models on their ability to adapt to temporal changes in social media flows. The results show significant model instability with accuracy drops reaching 23.4% during event-driven periods. The author proposes four new drift metrics validated on 12,279 authentic social media posts, achieving promising results for production deployment.

→

Dec 25 2025

LLM

Adversarial Training for Failure-Sensitive User Simulation in Mental Health Dialogue Optimization

A new approach for creating more realistic user simulators that enhance the safety and effectiveness of mental health support chatbots.

→

Dec 25 2025

LLM

SA-DiffuSeq: Addressing Computational and Scalability Challenges in Long-Document Generation with Sparse Attention

The X company has announced today the launch of SA-DiffuSeq, a new approach for long document generation that addresses computational and scalability challenges. The new framework integrates sparse attention to improve sampling efficiency and precision in long-range dependency modeling.

→

Dec 25 2025

LLM

La rivoluzione dei modelli neurali con meno parametri

Un nuovo approccio per i modelli neurali controllati differenziali (Neural CDEs) potrebbe rivoluzionare il campo dell'intelligenza artificiale. Questo metodo, che richiede molto meno parametri rispetto agli attuali modelli, offre una soluzione innovativa per analizzare sequenze temporali.

→

Dec 25 2025

LLM

Models of Large Language: A New Trail for Pedagogical Quality in Mathematics?

A recent study examined the behavior of large language models in mathematics education compared to expert human tutors. The results show that these models have a similar level of pedagogical quality as experts but use different teaching strategies and linguistic approaches.

→

Dec 25 2025

LLM

The Secret of LLM Models: Uncovering How Tokenizers Affect Their Performance

A new platform, TokSuite, has been created to study the role of tokenizers in improving LLM models. This technology allows researchers to delve deeper into the impact of tokenizers on model performance.

→

Dec 25 2025

LLM

New framework for detecting and mitigating spurious forgetting in continual learning

Recently discovered spurious forgetting is a fundamental challenge for language models. Continual learning is a technique that enables models to adapt to new information, but spurious forgetting can lead to performance degradation.

→

Dec 25 2025

LLM

The Future Transformation of Trains: Observe, Predict, and Learn

This article explores the future development of trains in Italy, discussing current trends and innovations that will shape the industry.

→

Dec 24 2025

Hardware

Nvidia licenses Groq's AI chip technology and hires its CEO

With the partnership with Groq, Nvidia becomes even more dominant in the chip manufacturing industry. Groq receives a financial investment from Nvidia to develop its AI chip technology.

→

Dec 24 2025

Altro

SK hynix expands US presence with new Seattle office

South Korean memory chip maker SK hynix has opened a new office in Seattle, expanding its US presence. The move is seen as a significant step forward for the company in strengthening ties with major clients.

→

Dec 24 2025

Altro

Artificial Intelligence Revolutionizes Operating Room Scheduling

The operating room is a complex and costly field that requires precise coordination between medical staff and scheduling. A startup believes artificial intelligence can improve this process, reducing costs and waiting times for patients.

→

Dec 24 2025

LLM

Why the operating room is ripe for AI, according to Akara

Hospitals are losing millions of euros to operating room inefficiencies, but AI is the key to solving complex coordination issues.

→

Dec 24 2025

LLM

Waymo tests Gemini as an in-car AI assistant for its robotaxis

Waymo is testing Gemini-powered in-car AI assistant for its robotaxis

→

Dec 24 2025

LLM

Italy tells Meta to suspend its policy that bans rival AI chatbots from WhatsApp

→

Dec 24 2025

LLM

China's reverse-engineered EUV chipmaking tool hasn't produced a single chip — sanctions-busting experiment is still years away from becoming operational

A reported attempt by a covert Chinese lab to reverse-engineer an EUV lithography scanner underscores that, despite access to scattered components, replicating ASML's EUV tools is effectively impossible without recreating the company's entire global supply chain, optics ecosystem, and proprietary software built over decades.

→

Dec 24 2025

LLM

Samsung delays DDR4 end-of-life due to long-term contract with key customer

→

Dec 24 2025

Market

New tariffs on Chinese electronic components announced by US administration

US announces new tariffs on Chinese electronic components, but implementation delayed until 2027

→

Dec 24 2025

LLM

How AI Code Agents Work: A Detailed Explanation

AI code agents, also known as large language models (LLMs), use neural networks to analyze vast amounts of data and complete code with a plausible response. They can be improved through fine-tuning and learning from human feedback.

→

Dec 24 2025

LLM

52 years later, only known copy of Unix v4 recovered from randomly found tape, now up and running on a system

A team of researchers has recovered the only known copy of Unix v4 from a tape found at the University of Utah. The operating system is now running on a platform.

→

Dec 24 2025

LLM

Efficient and Equitable Last-Mile Relief Aid Distribution

A new optimization algorithm can overcome limited resources to ensure effective and rapid humanitarian aid.

→

Dec 24 2025

LLM

Discovering Lie Groups with Flow Matching

The research proposes a new approach for discovering symmetries in data, improving performance and efficiency of machine learning models. The method, called \lieflow, uses flow matching on Lie groups to explore symmetries directly from data.

→

Dec 24 2025

LLM

FGDCC: Fine-Grained Deep Cluster Categorization -- A Framework for Intra-Class Variability Problems in Plant Classification

A team of researchers has developed a new algorithm to improve plant analysis. The method, known as FGDCC, uses classification to overcome the obstacles in representing images within a category. This work can open up new possibilities for developing more sophisticated machine learning models.

→

Dec 24 2025

LLM

Reducing Label Dependency in Human Activity Recognition with Wearables: From Supervised Learning to Novel Weakly Self-Supervised Approaches

A new study presents innovative solutions for human activity recognition with wearables, reducing label dependency. Researchers developed a weakly supervised framework that optimizes performance with only 10% of the labels.

→

Dec 24 2025

Frameworks

New technology to measure rhetorical style

A new LLM-based platform measures rhetorical style independently of substantive content.

→

Dec 24 2025

LLM

Hierarchical Agentic Reasoning for Multimodal Oncology Note Extraction

A new approach for extracting clinical data from oncology notes using large language models

→

Dec 24 2025

LLM

Bias Beneath the Tone: Empirical Characterization of Tone Bias in LLM-Driven UX Systems

Recently discovered the hidden biases in conversations with technology based on language models. A team of researchers analyzed language models and found that they can have tone tendencies, influencing user perception of trust, empathy, and fairness.

→

Dec 24 2025

LLM

Advanced AI models for predicting mortality of ICU patients

A recent research team has made a significant improvement to AI models for predicting mortality of ICU patients by integrating structured and unstructured clinical data.

→

Dec 24 2025

LLM

Revolution in Plant Species Classification with Open Techniques

A research team has developed a new method for multi-label classification of plant species on high-resolution images, achieving fifth place in the PlantCLEF 2025 challenge.

→

Dec 24 2025

Frameworks

PRISM: A Personality-Driven Multi-Agent Framework for Social Media Simulation

→

Dec 24 2025

LLM

PhysMaster: The Artificial Intelligence Revolutionizing Theoretical Physics and Computational Science

A new research project proposes using PhysMaster as an autonomous agency for theoretical physics and computational science, leveraging its advanced language model capabilities to accelerate research, automate tasks, and uncover independent discoveries.

→

Dec 24 2025

Frameworks

Synthetic Data Blueprint: un framework innovativo per la valutazione dei dati sintetici

Synthetic data are widely used in the rapidly evolving field of Artificial Intelligence to accelerate innovation while preserving privacy and enabling broader data accessibility. However, the evaluation of synthetic data remains fragmented across heterogeneous metrics, ad-hoc scripts, and incomplete reporting practices.

→

Dec 24 2025

LLM

Come i modelli Llama riconoscono i movimenti istruzioneali

Gli sviluppatori hanno valutato la capacità dei modelli Llama a riconoscere i movimenti istruzioneali nei testi autentici, scoprendo che solo con l'adeguamento del codice è possibile superare i limiti delle applicazioni di base.

→

Dec 24 2025

LLM

Thermodynamic Focusing for Inference-Time Search: Practical Methods for Target-Conditioned Sampling and Prompted Inference

Finding rare but useful solutions in very large candidate spaces is a recurring practical challenge across language generation, planning, and reinforcement learning. A new approach for more effective results.

→

Dec 24 2025

LLM

New Turn for Llama Models in EDA Sector

A new framework utilizing large language models to tackle the complex EDA sector has been developed. The solution combines fine-tuning of LLMs with text-to-text regression to significantly improve output format reliability.

→

Dec 23 2025

Altro

ASRock X870 Taichi Creator: The Perfect Choice for DIY Enthusiasts

The ASRock X870 Taichi Creator is a new arrival in the market of mid-range motherboards that proposes itself as the perfect choice for DIY enthusiasts. It features creator-focused characteristics, such as 10 GbE ports, ample storage slots, and two PCIe 5.0 slots, along with numerous USB ports.

→

Dec 23 2025

Altro

Authors Bring New Lawsuit Against Six Major AI Companies

Authors Reject Class Action Settlement with Anthropic, Arguing AI Companies Can't Easily Dismiss Thousands of Claims at Low Rates

→

Dec 23 2025

LLM

World’s largest shadow library made a 300TB copy of Spotify’s most streamed songs

The world's largest shadow library made a 300TB copy of Spotify’s most streamed songs, representing over 99% of listens. This massive dataset was distributed via torrents and is the first 'preservation archive' for music that is fully open.

→

Dec 23 2025

Altro

Nvidia prepares shipment of 82,000 AI GPUs to China as chip war lines blur

→

Dec 23 2025

LLM

YouTube bans two popular channels that created fake AI movie trailers

Two popular YouTube channels with millions of subscribers have been shut down for publishing fake AI movie trailers. The platform has applied content rules and forced the creators to clearly state that it was not official trailers.

→

Dec 23 2025

Hardware

Gigabyte removes controversial leaking thermal gel from RTX 5070 Ti Windforce V2

The second iteration of the Gigabyte RTX 5070 Ti triple-fan Windforce graphics card has switched from controversial leaking thermal gel to traditional thermal pads.

→

Dec 23 2025

LLM

Explainable Machine Learning Outperforms Linear Regression for Predicting County-Level Lung Cancer Mortality Rates in the United States

→

Dec 23 2025

Hardware

Legacy AMD GPUs receive 30% performance boost in Linux with new drivers

Linux 6.19 drops legacy Radeon DRM driver for the modern AMDGPU kernel driver, ushering performance uplifts up to 30% for AMD GCN 1.0 and GCN 1.1 graphics cards

→

Dec 23 2025

LLM

New Turn for Llama Models...

The technology of Artificial Intelligence (AI) is changing the face of marketing, enabling service agencies to offer more effective and faster solutions to their clients.

→

Dec 23 2025

LLM

Copilot for Italian Businesses: A New Step Towards Artificial Intelligence

Microsoft Copilot is a new tool that supports businesses in Italy, enabling them to leverage artificial intelligence in their workflow. With its integration with Microsoft 365 tools, it provides a default solution for employees working on consulting, delivery, management, and software development.

→

Dec 23 2025

LLM

Nuova svolta per i modelli Llama, self-play e formalizzazione

Un nuovo framework utilizza la formalizzazione dei codici per fornire segnali di correttezza ai modelli di linguaggio. I risultati sono promettenti, con un miglioramento del 96% rispetto ai basi di confronto.

→

Dec 23 2025

LLM

Child exploitation reports on OpenAI increased sharply this year

Child exploitation reports on OpenAI increased sharply this year according to a recent update from the company.

→

Dec 23 2025

Altro

Marissa Mayer’s new startup Dazzle raises $8M led by Forerunner’s Kirsten Green

→

Dec 23 2025

LLM

Supplementary Resources and Analysis for Automatic Speech Recognition Systems Trained on the Loquacious Dataset

The recently published Loquacious dataset aims to be a replacement for established English automatic speech recognition (ASR) datasets such as LibriSpeech or TED-Lium. The main goal of the Loquacious dataset is to provide properly defined training and test partitions across many acoustic and language domains, with an open license suitable for both academia and industry.

→

Dec 23 2025

Altro

Revolutionary Model for Traffic Flow Imputation: HINT Surpasses Standards

A new model called HINT (Hybrid Inductive-Transductive Network) has been presented, which combines inductive and transductive techniques to impute traffic flow in unmeasured locations.

→

Dec 23 2025

Altro

Intel Challenges AMD with Xeon Granite Rapids Workstation Lineup Featuring 86 Core Processors

Intel is set to launch a new line of Xeon Granite Rapids workstations, boasting 86 core processors. This move may shake up the workstation market, particularly as it competes directly with AMD's Threadripper models.

→

Dec 23 2025

LLM

New method for solving facility layout problems: a breakthrough for combinatorial optimization models

A new research has presented an innovative method for solving facility layout problems using combinatorial optimization techniques. The method, called Conflict-Driven Clause Learning with VSIDS heuristics, shows promising results for solving large-scale facility layout problems in near-constant time.

→

Dec 23 2025

LLM

Google's year in review: 8 areas with research breakthroughs in 2025

→

Dec 23 2025

LLM

Parameter-Efficient Fine-Tuning for HAR: Integrating LoRA and QLoRA into Transformer Models

Human Activity Recognition is a foundational task in pervasive computing. Recent advances in self-supervised learning and transformer-based architectures have significantly improved HAR performance, but adapting large pre-trained models to new domains remains a practical challenge due to limited computational resources.

→

Dec 23 2025

Altro

Convolutional-neural-operator-based transfer learning for solving PDEs

Innovative research offering a revolutionary solution for solving differential equations, improving precision and reducing computation time.

→

Dec 23 2025

LLM

The Revolution of Image Falsification: How OpenAI is Changing the Game

OpenAI's new image integration model, ChatGPT Image Generator, allows for easy creation of fake images using AI. This represents a significant advancement in image manipulation and has the potential to disrupt the photography industry.

→

Dec 23 2025

LLM

OpenAI warns AI browsers may always be vulnerable to prompt injection attacks

→

Dec 23 2025

Hardware

Chinese fabs upgrade older ASML DUV lithography chipmaking machines

Chinese fabs extend the useful life and performance of older ASML deep ultraviolet lithography systems by upgrading key subsystems

→

Dec 23 2025

LLM

USA minacciano sanzioni contro aziende europee, esperti prevedono una 'spia segreta' per l'UE

La US Trade Representative ha lanciato un attacco alle aziende europee accusate di discriminazione e molestie nei confronti dei provider americani. Ma suonerebbe all'altro posto se la critica europea dovesse avere un effetto positivo sull'economia globale?

→

Dec 23 2025

LLM

Lemon Slice nabs $10.5M from YC and Matrix to build out its digital avatar tech

Digital avatar generation company Lemon Slice is working to add a video layer to AI chatbots with a new diffusion model that can create digital avatars from a single image.

→

Dec 23 2025

LLM

5 ways AI agents will transform the way we work in 2026

Google Cloud's 2026 AI Agent Trends Report predicts that AI agents will revolutionize the way we work. This article explores five ways in which AI technology will be transformed to change our working lives.

→

Dec 23 2025

Altro

Efficient Mixture-of-Agents Serving via Tree-Structured Routing, Adaptive Pruning, and Dependency-Aware Prefill-Decode Overlap

→

Dec 23 2025

LLM

Zara Uses Artificial Intelligence to Speed Up Production Processes

Zara is using artificial intelligence to generate new model images in various outfits, based on past photos, without repeating costly photo shoots.

→

Dec 23 2025

LLM

Alexa+ integrates with new services: what's behind its AI power

Amazon announces Alexa+ can now integrate with new services: what's behind its AI power

→

Dec 23 2025

LLM

Inside China's push to apply AI across its energy system

China applies AI to manage energy system complexity, aiming for a more sustainable future.

→

Dec 23 2025

Hardware

Another unreleased Nvidia RTX 3080 Ti 20GB Founders Edition engineering sample surfaces on the second-hand market — a previous sample sold for $1,999 on eBay earlier this year

An engineering sample of the Nvidia RTX 3080 Ti 20GB has surfaced on the second-hand market, according to a post on the Nvidia subreddit.

→

🗄️ News Archive