Facebook Instagram Twitter Youtube
    Latest

    Web3

    Home Web3 Page 83

    MiniCPM 4.1-8B: Deploying Ultra-Efficient LLM on Spheron’s GPU Network

    By
    Metaverse Fashion Week Online
    -
    0
    MiniCPM 4.1-8B: Deploying Ultra-Efficient LLM on Spheron’s GPU Network


    For years, powerful AI models needed massive data centers and expensive cloud subscriptions. Now that’s changing. MiniCPM 4.1-8B is a new AI model that runs on regular computers and consumer GPUs. It performs as well as much larger models but uses far fewer resources.

    Think of it this way: instead of renting a semi-truck to move your furniture, you now have a compact van that does the same job faster and cheaper.

    What Makes MiniCPM 4.1-8B Special?

    MiniCPM 4.1-8B is an 8-billion-parameter language model that you can run on your own hardware. The team at OpenBMB built it from the ground up to be efficient.

    Four Key Innovations

    1. Smart Attention System (InfLLM v2)

    Most AI models read every single word when processing text. MiniCPM 4.1 skips this. It uses “sparse attention” to focus only on the most relevant parts of the text. Imagine reading a 500-page book but only highlighting the important paragraphs; that’s what InfLLM v2 does. It ignores 81% of the text while still understanding everything perfectly.

    2. Better Training Data

    The team trained MiniCPM 4.1 on just 8 trillion tokens of high-quality data. Compare this to Qwen3-8B, which needed 36 trillion tokens to reach similar performance. MiniCPM achieves the same results with just 22% of the training data. They filtered out low-quality content and generated reasoning-intensive data specifically for math and coding tasks.

    3. Two Modes: Fast and Deep

    You can run MiniCPM 4.1 in two ways:

    Fast mode: Quick responses for simple questions

    Deep reasoning mode: Detailed, step-by-step thinking for complex problems

    This flexibility lets you choose speed or depth based on your needs.

    4. Incredible Speed

    MiniCPM 4.1 processes long documents 7 times faster than Qwen3-8B on edge devices. When handling 128,000 words, it maintains this speed advantage throughout.

    Real Performance Numbers

    Here’s how MiniCPM 4.1-8B performs:

    General Knowledge: Scores 75-81% on major benchmarks (MMLU, CMMLU, CEval)

    Math Problems: Solves 91.5% of grade-school math correctly (GSM8K)

    Code Writing: Passes 85% of coding tests (HumanEval)

    Reasoning Tasks: Achieves 76.73% on complex reasoning (BBH)

    These scores match or beat models with twice as many parameters.

    How to Run MiniCPM 4.1 on Spheron Network

    Spheron Network gives you access to powerful GPUs without using traditional cloud providers like AWS or Google. You rent GPUs directly from providers worldwide. Let us walk you through the setup.

    Step-by-Step Setup Guide

    Step 1: Access Spheron Console and Add Credits

    Head over to console.spheron.network and log in to your account. If you don’t have an account yet, create one by signing up with your Email/Google/Discord/GitHub.

    Once logged in, navigate to the Deposit section. You’ll see two payment options:

    SPON Token: This is the native token of Spheron Network. When you deposit with SPON, you unlock the full power of the ecosystem. SPON credits can be used on both:

    Community GPUs: Lower-cost GPU resources powered by community Fizz Nodes (personal machines and home setups)

    Secure GPUs: Data center-grade GPU providers offering enterprise reliability

    USD Credits: With USD deposits, you can deploy only on Secure GPUs. Community GPUs are not available with USD deposits.

    For running NeuTTS, we recommend starting with Secure GPUs to ensure consistent performance. Add sufficient credits to your account based on your expected usage.

    Step 2: Navigate to GPU Marketplace

    After adding credits, click on Marketplace. Here you’ll see two main categories:

    Secure GPUs: These run on data center-grade providers with enterprise SLAs, high uptime guarantees, and consistent performance. Ideal for production workloads and applications that require reliability.

    Community GPUs: These run on community Fizz Nodes, essentially personal machines contributed by community members. They’re significantly cheaper than Secure GPUs but may have variable availability and performance.

    For this tutorial, we’ll use Secure GPUs to ensure smooth installation and optimal performance.

    Step 3: Search and Select Your GPU

    You can search for GPUs by:

    Region: Find GPUs geographically close to your users

    Address: Search by specific provider addresses

    Name: Filter by GPU model (RTX 4090, A100, etc.)

    For this demo, we’ll select a Secure RTX 4090 (or A6000 GPU), which has excellent performance for running NeuTTS. The 4090 provides the perfect balance of cost and capability for both testing and moderate production workloads.

    Click Rent Now on your selected GPU to proceed to configuration.

    Step 4: Select Custom Image Template

    After clicking Rent Now, you’ll see the Rent Confirmation dialog. This screen shows all the configuration options for your GPU deployment. Let’s configure each section. Unlike pre-built application templates, running NeuTTS requires a customized environment for development capabilities. Select the configuration as shown in the image below and click “Confirm” to deploy.

    GPU Type: The screen displays your selected GPU (RTX 4090 in the image) with specifications: Storage, CPU Cores, RAM.

    GPU Count: Use the + and – buttons to adjust the number of GPUs. For this tutorial, keep it at 1 GPU for cost efficiency.

    Select Template: Click the dropdown that shows “Ubuntu 24” and look for template options. For running NeuTTS, we need an Ubuntu-based template with SSH enabled. You’ll notice the template shows an SSH-enabled badge, which is essential for accessing your instance via terminal. Select: Ubuntu 24 or Ubuntu 22 (both work perfectly)

    Duration: Set how long you want to rent the GPU. The dropdown shows options like: 1hr (good for quick testing), 8hr, 24hr, or longer for production use. For this tutorial, select 1 hour initially. You can always extend the duration later if needed.

    Select SSH Key: Click the dropdown to choose your SSH key for secure authentication. If you haven’t added an SSH key yet, you’ll see a message to create one.

    Expose Ports: This section allows you to expose specific ports from your deployment. For basic command-line access, you can leave this empty. If you plan to run web services or Jupyter notebooks, you can add ports.

    Provider Details: The screen shows provider information:

    This shows which decentralized provider will host your GPU instance.

    Scroll down to the Choose Payment section. Select your preferred payment option:

    USD – Pay with traditional currency (credit card or other USD payment methods)

    SPON: Pay with Spheron’s native token for potential discounts and access to both Community and Secure GPUs

    The dropdown shows “USD” in the example, but you can switch to SPON if you have tokens deposited.

    Step 5: Check the “Deployment in Progress“

    Next, you’ll see a live status window showing every step of what’s happening, like: Validating configuration, Checking balance, Creating order, Waiting for bids, Accepting a bid, Sending manifest, and finally, Lease Created Successfully. Once this is complete, your Ubuntu server is live!

    Deployment typically completes in under 60 seconds. Once you see “Lease Created Successfully,” your Ubuntu server with GPU access is live and ready to use!

    Step 6: Access Your Deployment

    Once deployment completes, navigate to the Overview tab in your Spheron console. You’ll see your deployment listed with:

    Status: Running

    Provider details: GPU location and specifications

    Connection information: SSH access details

    Port mappings: Any exposed services

    Step 7: Connect via SSH

    Click the SSH tab, and you will see the steps on how to connect your terminal via SSH to your deployment details. It will look something like the image below, follow it:

    ssh -i -p root@

    Open your terminal and paste this command. Upon your first connection, you’ll see a security prompt requesting that you verify the server’s fingerprint. Type “yes” to continue. You’re now connected to your GPU-powered virtual machine on the Spheron decentralized network.

    Step 8: Install Miniconda

    We’ll install Miniconda to manage Python environments cleanly.This will make it easier to isolate dependencies for MiniCPM.

    wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh

    Run the installer silently (no prompts):

    bash ~/miniconda.sh -b -p ~/miniconda

    Initialize conda for bash:

    ~/miniconda/bin/conda init bash

    Step 9: Create and Activate the Conda Environment

    We’ll now create a new environment for MiniCPM and activate it, and Reloadthe shell so conda works right away:

    source ~/.bashrc
    conda create -n minicpm python=3.11 -y && conda activate minicpm

    Accept Conda’s Terms of Service to avoid setup interruptions:

    conda tos accept –override-channels –channel https://repo.anaconda.com/pkgs/main
    conda tos accept –override-channels –channel https://repo.anaconda.com/pkgs/r

    Recreate and activate just to make sure:

    conda create -n minicpm python=3.11 -y && conda activate minicpm

    If conda path issues appear, use this:

    source /root/miniconda/etc/profile.d/conda.sh && conda activate

    Step 10: Install Dependencies

    Now we’ll install all necessary packages, PyTorch, transformers, accelerate, and a few utilities.

    Install GPU-enabled PyTorch (CUDA 12.1):

    pip install torch>=2.0.0 –index-url https://download.pytorch.org/whl/cu121

    Install build tools and libraries:

    pip install “ninja>=1.0.0”
    pip install transformers
    pip install accelerate==0.26.0
    pip install –upgrade pip setuptools wheel
    pip install –upgrade aiohttp

    Step 11: Install Git and Clone the CPM.cu Repo

    We’ll now clone the OpenBMB CPM.cu repository, which contains the custom CUDA inference backend for MiniCPM models.

    apt update && apt install -y git

    Clone the repo (with submodules):

    git clone https://github.com/OpenBMB/CPM.cu.git –recursive && cd CPM.cu

    Step 12: Set Up CUDA and Build CPM.cu

    We’ll install CUDA Toolkit and build the CPM.cu backend.

    Install CUDA toolkit:

    conda install -c conda-forge cuda-toolkit -y

    Set the CUDA environment path, Build and install CPM.cu:

    export CUDA_HOME=/root/miniconda
    python3 setup.py install

    Step 13: Log in to Hugging Face

    You need to authenticate to download MiniCPM model weights.This opens a Hugging Face login prompt.

    When prompted, paste your Hugging Face access token. If you don’t have a token yet:

    Visit huggingface.co/settings/tokens

    Click “New token”

    Select “Read” permissions (sufficient for downloading models)

    Name it something memorable like “MiniCPM4.1”

    Copy the token and paste it when the terminal prompts you

    After successful authentication, you’ll see a confirmation message.

    hf auth login

    Step 14: Install the CPM.cu Python Package

    Make sure the package is installed properly so Python can import it.

    cd /root/CPM.cu && pip install .

    Step 15: Connecting a Code Editor

    Connect your GPU VM by running the same command you have used to connect your GPU in the terminal.

    ssh -i -p root@

    Now go to the CPM.cu folder > examples > Create a file named prompt.txt. In prompt.txt, you can add your prompt, which you want to run through MiniCPM 4.1. Save the file and go back to the terminal.

    Step 16: Run the MiniCPM Inference Demo

    Now, everything’s ready. Let’s test MiniCPM 4.1-8B with a sample prompt.This runs the example inference script included in CPM.cu.

    python3 /root/CPM.cu/examples/minicpm4/test_generate.py –prompt-file /root/CPM.cu/examples/prompt.txt

    This will load the MiniCPM model, generate text for the prompt, and print results in the terminal.

    You’ve successfully deployed MiniCPM 4.1-8B on a Spheron decentralized GPU. You now have:

    A fully local, private inference environment

    A lightweight, efficient LLM runtime

    Access to the CPM.cu CUDA backend for max GPU efficiency.

    Conclusion

    MiniCPM-4.1-8B proves that efficiency and power can go hand in hand, delivering state-of-the-art performance through innovations in architecture, training, data, and inference while remaining lightweight enough for local or GPU-based deployment. With the help of CPM.cu, users can unlock the model’s full potential by leveraging optimized sparse attention, quantization, and CUDA-based acceleration. Spheron Network makes this entire journey seamless by providing decentralized, cost-efficient GPU infrastructure, simplifying deployment, scaling, and environment management. Developers can now focus on rapid experimentation and results with pre-configured, GPU-powered by Spheron’s global compute network.



    Source link

    Class Action Alleges Microsoft Choked AI Supply to Lift ChatGPT Costs – Decrypt

    By
    Metaverse Fashion Week Online
    -
    0
    Class Action Alleges Microsoft Choked AI Supply to Lift ChatGPT Costs – Decrypt



    In brief

    Microsoft allegedly used an exclusive Azure agreement from 2019 to control compute supply to OpenAI, keeping ChatGPT prices artificially high while developing competing AI products.
    When OpenAI began buying compute from Google in June 2025, token prices reportedly dropped 80 percent within weeks, which plaintiffs say is evidence of the restraint.
    The lawsuit seeks damages for overcharges from November 2022 to February 2025, warning Microsoft still holds contractual power to throttle OpenAI’s supply.

    ChatGPT users have accused Microsoft of “mercilessly” choking OpenAI’s compute supply through an exclusive cloud agreement, artificially inflating AI subscription prices while simultaneously rushing its own competing products to market.

    The lawsuit, filed Monday in San Francisco federal court, alleges Microsoft “secretly turned an investment into a stranglehold,” using its Azure cloud dominance to restrict the computational resources needed to run ChatGPT, keeping prices at levels reaching “100 to 200 times” competitors’ rates during a February 2025 AI price war.

    Eleven ChatGPT Plus subscribers brought the case, alleging they overpaid for subscriptions and received degraded service from November 2022 through February 2025 due to Microsoft’s anticompetitive conduct.

    

    The lawsuit calls the case “a sequel to U.S. v. Microsoft,” invoking the company’s 1990s antitrust battles and characterizing it as a “recidivist violator” that has “ported the same exclusionary playbook into AI.”

    Microsoft’s 2019 deal gave exclusive rights to supply Azure compute to OpenAI’s commercial models, effectively giving the company contractual control over a “horizontal competitor’s supply chain” while it built its own generative AI products, according to the filing.

    The plaintiffs define a new antitrust market, the Consumer Generative AI Market, encompassing subscription products like ChatGPT Plus, Claude Pro, Gemini Advanced, and DeepSeek Chat. 

    Microsoft reportedly holds a 49% stake in OpenAI’s for-profit arm and takes 20% of its paid-product revenue, the lawsuit alleges, effectively “profiting twice—first from compute sales, and again from the very AI product it constrains.”

    When OpenAI began purchasing compute from Google Cloud in June 2025, ending Microsoft’s exclusivity, ChatGPT token prices dropped 80 percent within weeks, according to the complaint.

    A ‘powerful experiment’

    The complaint calls it a “powerful natural experiment” showing Microsoft’s “anticompetitive restraint,” noting that before the shift, ChatGPT users faced “poor quality, unreleased innovations, and slow response times.”

    “On proof, the strongest evidence would be the exclusive agreement itself,” Navodaya Singh Rajpurohit, legal partner at Coinque Consulting, told Decrypt. “If that document shows Microsoft held or used control over OpenAI compute, that is primary proof. If it is not available, internal emails and capacity records can still carry the case.”

    “Restraint of trade turns on control,” Rajpurohit said, adding the claim is strong if Microsoft actually exercised or clearly held that control, but weaker if it did not.

    The complaint also alleges Microsoft “still retains the contractual ability to restrict OpenAI’s compute purchases,” a power that lingers “as a sword of Damocles over OpenAI, wielded by one of its principal competitors.”

    Plaintiffs are seeking monetary damages and a permanent injunction barring Microsoft from exclusive compute deals with OpenAI, along with disclosure of internal communications on compute supply, pricing, and integration.

    “The court can do more than fine the company,” Rajpurohit explained. “It can order changes to the OpenAI arrangement, remove exclusive terms, and set guardrails that prevent future choke points.”

    The class-action arrives amid shifts in the Microsoft-OpenAI relationship. Japan’s SoftBank is reportedly negotiating to invest up to $25 billion in OpenAI, surpassing Microsoft’s $13 billion stake. Meanwhile, Microsoft has agreed under the Stargate project to end its exclusive cloud provider status.

    Microsoft and OpenAI did not immediately respond to Decrypt’s requests for comment.

    Generally Intelligent Newsletter

    A weekly AI journey narrated by Gen, a generative AI model.



    Source link

    Heat Therapy Is Becoming the Next Big Thing in Recovery with Nordica Sauna | Web3Wire

    By
    Metaverse Fashion Week Online
    -
    0
    Heat Therapy Is Becoming the Next Big Thing in Recovery with Nordica Sauna | Web3Wire


    Image: https://www.abnewswire.com/upload/2025/10/919d8a46f94d3883cefe37253cd0f50e.jpg

    As we explore innovative approaches to recovery, heat therapy has emerged as a front-runner in promoting healing and well-being. This ancient technique is reclaiming its place in modern science, supported by a growing body of research. Brands like Nordica Sauna [https://nordicasauna.com/] exemplify how traditional thermal therapy can be reimagined through modern design, providing accessible wellness solutions for everyday use.

    Studies from reputable sources such as the Mayo Clinic highlight its effectiveness in recovery strategies. In this text, we will investigate deep into the science, benefits, and practical applications of heat therapy, demonstrating why it is becoming essential for anyone interested in optimizing their recovery process.

    Understanding Heat Therapy

    The Science Behind Heat Therapy

    Heat therapy, also called thermotherapy, leverages the body’s response to warmth to help recovery. When heat is applied, blood vessels dilate, enhancing blood flow and delivering vital nutrients to the affected tissues. This process aids in muscle relaxation, pain relief, and accelerated healing.

    Types Of Heat Therapy

    Heat therapy can be classified into two primary categories: dry heat and moist heat. Dry heat includes methods like heating pads and infrared saunas, which deliver heat without moisture. Conversely, moist heat methods, such as warm towels or hydrotherapy, use water to provide therapeutic warmth. Understanding these types allows us to choose the right approach based on our recovery needs.

    Benefits Of Heat Therapy In Recovery

    Enhancing Blood Circulation

    One of the most significant benefits of heat therapy is its ability to promote better blood circulation. Increased circulation ensures oxygen and nutrients are efficiently delivered to tissues, facilitating repair. Many athletes incorporate this method post-exercise to expedite recovery.

    Promoting Muscle Relaxation

    Heat naturally relaxes our muscles, reducing tension and stiffness. After an intense workout or injury, applying heat helps to soothe the muscles, making them more pliable and ready for rehabilitation exercises. This property is particularly beneficial for chronic conditions like arthritis.

    Reducing Pain And Stiffness

    We often face lingering pain and stiffness from various activities or injuries. Heat therapy effectively alleviates these discomforts by soothing sore muscles and reducing muscle spasms. According to a study published by the American Academy of Family Physicians, participants reported significant pain relief after consistent use of heat therapy.

    Heat Therapy Techniques And Methods

    Hot Packs And Heating Pads

    Hot packs and heating pads are among the most accessible and popular forms of heat therapy. Placed on targeted areas, they provide localized warmth that penetrates deep into the muscles, enhancing recovery effects. For ease of use, electric heating pads often come with adjustable heat settings.

    Infrared Saunas

    Infrared saunas take heat therapy to the next level. They use infrared light to warm the body directly instead of the surrounding air, promoting deeper tissue penetration. Sessions can help detoxify, boost metabolism, and relieve muscle tension, making it an excellent addition to any recovery routine.

    Warm Baths And Hydrotherapy

    Soaking in warm baths is not only luxurious but also beneficial for recovery. Hydrotherapy, using water’s buoyancy, supports the body and reduces joint stress, all while delivering therapeutic warmth. Adding Epsom salts can enhance the muscle-relaxing effects.

    Heat Therapy In Sports Recovery

    Professional Athletes And Heat Therapy

    Many professional athletes have turned to heat therapy as a cornerstone of their recovery regimen. Teams often use heat therapy to speed up recovery processes, especially during intense training or after games. We see this in practices where athletes use heating pads after strenuous workouts to mitigate post-workout soreness.

    Integration Into Rehabilitation Programs

    Physical therapists increasingly integrate heat therapy into rehabilitation programs. It acts as a preparatory step before stretching and strength training, enhancing flexibility and overall performance. Utilizing heat enables patients to achieve better outcomes in their recovery.

    Safety Precautions And Considerations

    Potential Risks Of Heat Therapy

    While heat therapy has numerous benefits, it is essential to use it correctly. Overuse can lead to burns, especially for individuals with sensory impairments. We must ensure we monitor heat levels and duration to avoid these risks.

    When To Avoid Heat Therapy

    Certain conditions may warrant avoiding heat therapy. For example, if we have acute injuries or swelling, applying cold therapy may be more effective. Also, individuals with diabetes or circulatory issues should consult healthcare professionals before use. According to the National Institutes of Health [https://www.nih.gov/], heat therapy acts by stimulating the sensory receptors in our skin, which can inhibit the transmission of pain signals to the brain.

    Media ContactCompany Name: Nordica SaunaContact Person: Mr.DavidEmail:Send Email [https://www.abnewswire.com/email_contact_us.php?pr=heat-therapy-is-becoming-the-next-big-thing-in-recovery-with-nordica-sauna]Address:9620 S Las Vegas Blvd, Suite E4City: Las VegasState: NV 89123Country: United StatesWebsite: http://nordicasauna.com

    Legal Disclaimer: Information contained on this page is provided by an independent third-party content provider. ABNewswire makes no warranties or responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you are affiliated with this article or have any complaints or copyright issues related to this article and would like it to be removed, please contact retract@swscontact.com

    This release was published on openPR.

    About Web3Wire Web3Wire – Information, news, press releases, events and research articles about Web3, Metaverse, Blockchain, Artificial Intelligence, Cryptocurrencies, Decentralized Finance, NFTs and Gaming. Visit Web3Wire for Web3 News and Events, Block3Wire for the latest Blockchain news and Meta3Wire to stay updated with Metaverse News.



    Source link

    Alleged ‘Trump Insider Whale’ Denies Insider Trading, Opens New $340 Million Bitcoin Short – Decrypt

    By
    Metaverse Fashion Week Online
    -
    0
    Alleged ‘Trump Insider Whale’ Denies Insider Trading, Opens New 0 Million Bitcoin Short – Decrypt



    In brief

    A whale trading on Hyperliquid that made millions shorting BTC and ETH last week is now shorting Bitcoin again.
    The trader added $40 million USDC to Hyperliquid on Monday, using it to short Bitcoin with 10x leverage.
    The open position is currently yielding around $700,000 in unrealized profits.

    A whale that earned close to $200 million by shorting Bitcoin and Ethereum prior to Trump’s tariff announcement on Friday—which helped lead to a record $19 billion in crypto liquidations—has taken another stand against the top crypto asset.

    And while the wallet owner has been accused of being a “Trump insider,” the person purportedly behind it insists that there’s no connection to the First Family.

    The Ethereum address ending in “7283ae” deposited $40 million in USDC to perpetuals trading platform and decentralized exchange Hyperliquid on Monday morning, according to data from Hyperliquid block explorer HypurrScan.

    Shortly thereafter, the account began building a 10x short position in Bitcoin valued around $340 million. In other words, the account holder is betting on the price of Bitcoin to go down, and using 10x leverage to increase the size of their bet without needing to commit the full $340 million principal.

    Based on 7283ae’s entry price of $116,009, the account has already amassed more than $700,000 in unrealized profits. The entire position will be liquidated, wiping out the principal and gains, if Bitcoin reaches a new all-time high of $130,460. 

    Crypto researchers have pointed to the uncanny timing in the account’s recent actions on Hyperunit—a platform that allows native tokens like Bitcoin and Ethereum to be deposited and ultimately traded on Hyperliquid—as reasons to suspect having inside knowledge of President Trump’s market-sinking comments.

    

    Blockchain data firm Arkham Intelligence has labeled the wallet holder a “Trump insider whale,” and other crypto commentators have flinged similar accusations at the trader, though there’s no direct evidence to prove prior knowledge of Trump’s actions.

    Data from Hypurrscan and Arkham shows that the same wallet deposited $80 million in USDC to Hyperliquid via Hyperunit on Friday. It then opened about 3,700 BTC—or around $450 million—in Bitcoin shorts according to data from Hypurrscan. 

    One day later, the wallet withdrew $150 million from Hyperliquid and later transferred it to a new wallet, which now houses around $386 million in USDC. 

    A report from blockchain researcher Conor Grogan tied the account to a Bitcoin whale that had swapped millions of BTC for ETH earlier this year. Others have suggested that the account belongs to former BitForex CEO Garrett Jin, and he confirmed that he’s connected to it—but said it’s his “clients’ fund,” not his own personal account.

    Pseudonymous on-chain sleuth “Eyeonchains” first flagged the connection between Jin and the unknown Bitcoin whale in a post on X over the weekend. The post was amplified by Binance founder Changpeng “CZ” Zhao, who reposted the allegation with the caption: “Not sure of validity. Hope someone can cross check.”

    Jin then wrote in reply this morning: “Hi CZ, thanks for sharing my personal and private information. To clarify, I have no connection with the Trump family or Donald Trump Jr.—this isn’t insider trading.”

    Bitcoin has risen modestly in the last 24 hours to $115,796, but still remains down 8% on the week following Friday’s market crash. Ethereum is up nearly 4% in the last 24 hours, and has fallen around 9% this week to a price of $4,284.

    Daily Debrief Newsletter

    Start every day with the top news stories right now, plus original features, a podcast, videos and more.





    Source link

    Top 9 Cloud GPU Providers for AI and Deep Learning in South Africa

    By
    Metaverse Fashion Week Online
    -
    0
    Top 9 Cloud GPU Providers for AI and Deep Learning in South Africa


    South Africa sits at the edge of a huge continental opportunity. We build for a market that demands cost efficiency, respects data sovereignty, and needs infrastructure that scales across different regions. Choosing a GPU cloud partner is more than a line item on a budget. It shapes your ability to innovate, ship, and compete.

    This guide maps practical choices for South African teams in 2025.

    1. Spheron AI

    Spheron AI fits teams that want bare-metal performance without enterprise complexity. It offers root-access VMs and bare-metal instances across an aggregated global network. That means you can deploy a GPU in minutes, tune drivers, and run heavy training jobs with no hypervisor overhead. For South African teams that need price predictability, Spheron keeps billing simple and removes common cloud surprises like hidden egress fees.

    If your priority is to run large models while keeping costs steady, Spheron is worth testing. It supports H100, A100, L40S, and a broad mix of consumer and datacenter GPUs. It integrates with Terraform and common MLOps tools, so you can automate provisioning without rewriting pipelines.

    Spheron also focuses on giving you a choice. When local capacity is limited, it aggregates providers so you don’t wait days for hardware. When you need tight control, you can pick full VMs and tune kernels. For South African teams balancing performance and budget, that flexibility reduces risk and speeds development.

    2. Nebius

    Nebius stands out for high-speed networking and automation. It gives you InfiniBand meshes and Terraform-friendly APIs. Use Nebius when you need low-latency, multi-node training across many GPUs.

    For teams working on large language models or multi-node vision jobs, Nebius reduces communication overhead between GPUs. That speeds throughput and often cuts total training time. The pricing is higher than basic spot marketplaces, but you pay for consistent performance and enterprise-grade automation.

    3. Lambda Labs

    Lambda Labs is engineered for researchers and engineering teams who want ready-made ML stacks and reliable multi-GPU clusters. They provide Lambda Stack images and 1-click cluster creation, which saves setup time for teams that want to run experiments right away.

    If you want a familiar environment and predictable multi-node performance, Lambda is a sensible choice. Their support for InfiniBand and tuned drivers makes it easier to move from prototype to sustained training runs.

    4. RunPod

    RunPod is flexible and developer-friendly. It supports serverless GPU endpoints and pod-based persistent instances. That hybrid model is great when you want to pay for compute only while code runs, but still need long-running pods for heavy jobs.

    Startups use RunPod for quick iterations, APIs, and cost-conscious inference. The per-second billing for serverless endpoints often lowers bills for bursty traffic. It also lets teams deploy custom Docker images quickly, which reduces friction when you want to test different stacks.

    Vast.ai is a marketplace that surfaces spare capacity from many hosts. It gives you extreme price flexibility. If your workloads tolerate interruptions or you want cheap batch training, Vast.ai can dramatically cut costs.

    The trade-off is consistency. Spot-like availability means you may see variable performance. But for many South African projects, early research, proof-of-concept training, and experimental hyperparameter sweeps Vast.ai gives you access to diverse hardware at deep discounts.

    focuses on fast provisioning and dynamic cost optimization. Their platform converts idle capacity into cheaper pools and offers serverless model APIs. Use it if you need to reserve capacity occasionally but also want cheap burst compute.

    Their environment includes preconfigured machine images and Kubernetes-native tooling. For teams that want strong automation and cost-sensitivity in one platform, it is a practical option.

    6. Genesis Cloud

    Genesis Cloud brings large-scale H100 and A100 clusters with a focus on sustainability and compliance. It is a good fit for enterprise teams that need sustained throughput, EU-compliant certs, and dense infrastructure for big training runs.

    If your workload needs consistent multi-node performance and you care about energy efficiency or regulatory certifications, Genesis Cloud gives a predictable, compliant option.

    7. Vultr

    Vultr provides a broad global footprint with many cost tiers. It offers a variety of GPUs, from consumer cards to powerful H100 variants. Vultr is useful when you need to place inference endpoints closer to end users.

    For teams with regional audiences or those that need multiple edge locations, Vultr’s many data centers reduce latency and give flexible deployment options. The pricing spectrum helps teams mix high-end training with low-cost inference where it makes sense.

    8. Gcore

    Gcore pairs GPU compute with an extensive global CDN and edge points. That makes it attractive for low-latency inference across continents. If you serve applications that must respond fast to users across Africa and Europe, Gcore’s edge reach reduces round-trip time and improves user experience.

    Gcore also has strong security features and enterprise tooling. Use it when you need to serve models at the edge while preserving control and compliance.

    9. OVHcloud

    OVHcloud offers dedicated GPU servers, hybrid options, and transparent pricing. It is known for single-tenant hardware, which helps when you need predictable performance and clear cost models.

    OVHcloud suits teams that require hybrid integrations with on-prem systems, or those that want straightforward capacity without the surprises of shared cloud layers.

    How to pick the right provider

    Start with requirements, not marketing. Ask: what matters most, raw price, low-latency inference, predictable multi-node throughput, or data residency? The answer drives the right choice.

    If price and flexibility dominate, test a marketplace or Spheron AI spot pools. If consistent multi-node training is critical, prioritize Spheron AI, Nebius, Lambda, or Genesis Cloud. If you need edge inference across countries, evaluate Gcore and Vultr for their CDN/edge reach. If you want a balanced, developer-friendly option with a lower price and full VM access, try Spheron AI.

    Always pilot with a real workload. Run a short training job that represents your production load. Measure throughput, GPU utilization, and actual wall-clock training time. Track network egress and storage charges. The numbers tell a different story than the advertised price per hour.

    Practical billing and FinOps tips

    Model your budget on dollars per useful throughput, not dollars per GPU hour. A GPU with better interconnect or higher sustained throughput can be cheaper in practice because it finishes jobs faster.

    Watch egress, snapshots, and cross-region transfers. Those network charges compound when you move large datasets. Prefer providers that bundle network or offer local storage to minimize surprise fees.

    Use reserved capacity for steady, predictable jobs. Use spot markets for burst and research tasks. Automate power-off for test VMs and use job queuing to avoid idle GPUs. One well-scripted FinOps change often slices 20%–40% off monthly cloud bills.

    Data sovereignty and compliance

    South African law and POPIA mean teams sometimes prefer local or regional hosting. If data residency matters, ensure the provider offers South African or nearby regional points of presence. For sensitive datasets, prefer single-tenant hardware or private VPCs. Confirm how providers handle backups, logs, and access control; those are often the gaps that create legal exposure.

    If you use aggregated networks, make sure you keep provenance records and clear contractual clauses on data use. Many platforms provide contractual guarantees that they won’t use your data to train models. Get that in writing if it matters to you.

    Performance checks to run during any trial

    Run a simple checklist before committing:

    Start a pilot with your real dataset.

    Measure GPU utilization and host overhead.

    Time a single training epoch and extrapolate cost to full runs.

    Test multi-node sync performance if you will scale horizontally.

    Check network throughput to your storage.

    Validate startup time and image boot times.

    Confirm snapshot and restore speed for disaster recovery.

    These checks reveal real costs, not marketing numbers. They also uncover hidden bottlenecks like slow S3-compatible endpoints or driver mismatches.

    Typical migration patterns

    Many South African teams use a hybrid approach. They keep sensitive workloads on dedicated hardware or local private clouds and shift training bursts to a GPU cloud. They run production inference on stable bare-metal providers and scale experiments on marketplaces or spot resources.

    This split reduces risk and preserves agility. It also lets teams capture the best per-use pricing and avoids vendor lock-in.

    When to negotiate and what to ask for

    If you plan sustained usage, ask providers about committed discounts, multi-month reservations, or dedicated racks. Negotiate for included egress, predictable network SLAs, and guaranteed availability windows during business hours.

    Ask for technical support SLAs and hands-on onboarding help. Often small credits for initial work or expert sessions speed your time-to-value.

    Final recommendation

    Start with a two-week pilot on the provider that best matches your primary constraint. Use a real training job and an inference test. Measure the total dollars spent, the actual throughput, and the engineering time required to keep the system healthy.

    If your primary concern is cost and you can tolerate interruptions, start with a marketplace like Spheron AI or spot pools. If you need multi-node performance, prioritize Nebius or Lambda. If you need predictable production throughput and lower overhead, try Spheron AI and test a bare-metal VM for a week.

    Infrastructure is not a solved problem. But the right choices make AI cheaper, faster, and simpler to operate. South African teams can win by matching their needs to the right provider, piloting early, and using a hybrid mix to balance price and reliability.



    Source link

    Ethereum-Based Mutuum Finance (MUTM) Records 62% Phase 6 Completion and $17.2M in Funding | Web3Wire

    By
    Metaverse Fashion Week Online
    -
    0
    Ethereum-Based Mutuum Finance (MUTM) Records 62% Phase 6 Completion and .2M in Funding | Web3Wire


    Mutuum Finance (MUTM) continues to build momentum as investor participation accelerates. The decentralized finance (DeFi) lending protocol has surpassed $17.2 million in total funding, with 62% of Phase 6 already sold out, signaling growing demand ahead of its planned launch. With a structured, transparent model and a clear development roadmap, Mutuum Finance is positioning itself as one of the most closely watched DeFi token launches of 2025.

    Detailed Presale Breakdown

    Mutuum Finance’s presale structure stands out in a market where many early-stage projects rely on vague timelines and unpredictable token allocations. The MUTM presale follows a fixed-stage pricing model designed to reward early participants with built-in appreciation at each phase.

    The sale began in Phase 1 at $0.01 per token, offering early backers one of the lowest entry points in this cycle. Since then, the price has climbed to $0.035 in Phase 6, representing a 250% increase for initial participants. Each phase raises the price by approximately 20% once the allocation is sold out, providing investors with full transparency on pricing progression.

    To date, over $17.2 million has been raised, more than 750 million tokens have been sold, and the investor base has expanded to over 16,900 holders. This growing participation reflects both retail and larger buyers taking positions as the project advances toward launch.

    Phase 6 has now reached 62% completion, with Phase 7 set to lift the token price to $0.04. The final listing price is locked at $0.06, meaning Phase 6 participants are still securing their tokens at a significant discount to the launch valuation. This structured approach is a notable departure from the unpredictable pricing strategies seen in many other token launches, and it has been instrumental in sustaining momentum throughout the campaign.

    Development Roadmap and X Statement

    Mutuum Finance is not relying on token sales alone. The team recently confirmed via an official X statement that Version 1 (V1) of its decentralized lending and borrowing protocol is on track to launch on Sepolia testnet in Q4 2025.

    This initial deployment will feature all the essential building blocks of the platform, including liquidity pools, debt tokens, a liquidator bot, and support for ETH and USDT as the first assets for lending, borrowing, and collateral. This is a crucial step in aligning product development with token distribution, something that many presales fail to do.

    By giving the community visibility into concrete milestones and introducing the testnet rollout, Mutuum Finance reduces reliance on speculation and provides investors with a clear, verifiable roadmap. This level of operational clarity has been one of the key reasons analysts are closely tracking MUTM’s progression through its later presale stages.

    Security, Transparency and Community Engagement

    Security has been a major focus for Mutuum Finance from the outset. The project underwent a CertiK audit, receiving a 90/100 Token Scan score, which indicates a strong technical foundation and well-structured smart contract architecture.

    To encourage independent testing, the team has also launched a $50,000 tiered bug bounty program, inviting external developers and security researchers to identify potential vulnerabilities ahead of mainnet deployment. This proactive approach not only strengthens protocol resilience but also signals the team’s commitment to transparency and best practices, qualities that are increasingly critical in the DeFi space.

    Community engagement is another cornerstone of the presale strategy. A $100,000 giveaway campaign is currently underway, rewarding ten participants with $10,000 worth of MUTM each. Additionally, a Top-50 leaderboard tracks the largest contributors during the presale, granting bonus allocations to those who rank highly. This gamified structure has kept participation levels high throughout the sale, while a dashboard allows investors to track their holdings, monitor ROI potential, and follow the progression of each presale phase transparently.

    Utility at the Core

    While this article primarily focuses on presale data, it’s worth briefly outlining the utility that underpins Mutuum Finance. MUTM is an Ethereum-based DeFi lending and borrowing protocol that aims to tie token demand directly to platform activity.

    The protocol features a dual-market structure that blends flexibility with risk management. In the Peer-to-Contract (P2C) market, major assets like ETH and USDT are supplied to shared liquidity pools, where lenders earn yield through variable interest rates while borrowers take overcollateralized loans against these assets, generating real economic activity within the system. 

    Meanwhile, the Peer-to-Peer (P2P) market supports riskier or lower-liquidity tokens such as DOGE or SHIB through isolated agreements between users, allowing the protocol to accommodate a wider range of assets without exposing its core liquidity pools to excessive volatility.

    Borrowing operates through overcollateralization — for example, depositing $10,000 in ETH allows a user to borrow up to $7,500 in stablecoins at a 75% Loan-to-Value (LTV) ratio. Lenders receive mtTokens, ERC-20 tokens minted 1:1 to their deposits (e.g., depositing 10,000 USDT mints 10,000 mtUSDT), which automatically accrue yield over time. This structure provides a clear economic engine behind MUTM, giving analysts tangible fundamentals to assess rather than relying solely on speculative narratives.

    Why These Milestones Matter

    Mutuum Finance’s presale performance underscores a growing trend: investors are increasingly drawn to projects that combine clear tokenomics, real utility, and transparent development roadmaps.

    With $17.2 million raised, 62% of Phase 6 completed, and a final listing price of $0.06 on the horizon, the project has entered a decisive stage. Its blend of technical progress, community engagement, and structured pricing distinguishes it from the many speculative token launches that dominate the early stages of bull cycles.

    As Mutuum Finance approaches its next pricing milestone and prepares for testnet deployment in Q4 2025, it is solidifying its position as one of the most closely followed DeFi projects heading into the new year. The coming months will determine how this early traction translates into real on-chain adoption — but for now, the numbers and the roadmap are aligned.

    For more information about Mutuum Finance (MUTM) visit the links below:

    Website: https://www.mutuum.com

    Linktree: https://linktr.ee/mutuumfinance

    Disclaimer: The information provided in this press release is not a solicitation for investment, nor is it intended as investment advice, financial advice, or trading advice. Investing involves risk, including the potential loss of capital. It is strongly recommended you practice due diligence, including consultation with a professional financial advisor, before investing in or trading cryptocurrency and securities. Neither the media platform nor the publisher shall be held responsible for any fraudulent activities, misrepresentations, or financial losses arising from the content of this press release.

    About Web3Wire Web3Wire – Information, news, press releases, events and research articles about Web3, Metaverse, Blockchain, Artificial Intelligence, Cryptocurrencies, Decentralized Finance, NFTs and Gaming. Visit Web3Wire for Web3 News and Events, Block3Wire for the latest Blockchain news and Meta3Wire to stay updated with Metaverse News.



    Source link

    MARA Holdings Buys $46 Million in Bitcoin Post-Crypto Market Tumble – Decrypt

    By
    Metaverse Fashion Week Online
    -
    0
    MARA Holdings Buys Million in Bitcoin Post-Crypto Market Tumble – Decrypt



    In brief

    MARA Holdings purchased 400 BTC worth $46.29 million from institutional liquidity provider FalconX.
    The purchase comes as Bitcoin rebounded to $114,763 after the largest liquidation event in crypto history wiped out over $19 billion in positions last Friday.
    Analysts say the acquisition points to the firm’s confidence that Bitcoin has “more room to run” as Trump softens tariff rhetoric and global monetary easing remains on the table.

    Bitcoin miner MARA Holdings snapped up 400 BTC worth $46.29 million from institutional crypto liquidity provider FalconX earlier today, as institutional investors view last week’s historic market crash as a buying opportunity rather than the start of prolonged weakness.

    The purchase, conducted through MARA’s wallet address “3MYao,” pushes the publicly-traded mining company’s total holdings to over 53,000 BTC, maintaining its position as second-largest corporate Bitcoin holder behind Strategy’s 640,031 BTC, according to Bitcoin Treasuries Net data.

    The acquisition comes as Bitcoin has rebounded to $114,763, up 3.2% in the last 24 hours, according to CoinGecko data, following what became the largest liquidation event in crypto history on Friday.

    Over $19 billion in crypto positions were wiped after President Donald Trump threatened “massive” tariffs against China, sending Bitcoin plummeting from above $121,000 to below $106,000 before recovering.

    Markets stabilized over the weekend after Trump softened his rhetoric, posting on Truth Social that Washington “wants to help China, not hurt it,” and describing Chinese President Xi Jinping as “highly respected.” 

    “The market broke down into chaos last week and almost immediately everybody was buying,” Pav Hundal, Lead Market Analyst at Swyftx, told Decrypt.

    “This was the largest liquidation event we’ve seen in crypto, but each time we see resets and the market just goes about its business again, which is exactly what seems to be happening with MARA,” Hundal added.

    

    Hundal said MARA appears to be “looking at the geo-economics and taking a call that Bitcoin now has more room to run,” noting potential for additional global monetary easing as “inflation forecasts are facing a double whammy at the moment with both oil prices and demand down.”

    MARA’s stock closed at $18.64 on October 10, down 7.75% from its previous close of $20.20, pointing to broader market weakness, according to Google Finance data.

    Daily Debrief Newsletter

    Start every day with the top news stories right now, plus original features, a podcast, videos and more.



    Source link

    BNB News: Sparkvia AI Launches $SPARK Presale on BNB Chain—Bringing The First AI-Powered Writing Platform to BNB Ecosystem | Web3Wire

    By
    Metaverse Fashion Week Online
    -
    0
    BNB News: Sparkvia AI Launches $SPARK Presale on BNB Chain—Bringing The First AI-Powered Writing Platform to BNB Ecosystem | Web3Wire


    VALLETTA, Malta, Oct. 12, 2025 (GLOBE NEWSWIRE) — Sparkvia AI announces the launch of the SPARK ($SPK) presale sale on BNB Chain, opening access to a utility token that fuels a pay-as-you-go credit system for AI writing.

    Positioned as the first AI-powered writing platform bringing a credit-based model to the BNB ecosystem, Sparkvia lets creators turn prompts into publish-ready content across websites, blogs, emails, product pages, and socials without subscriptions or tier lock-ins.

    Inside Sparkvia AI platform, $SPK Token will be used to purchase Spark credits, the single token that powers every prompt’s generation, at a predictable, equal cost per prompt.

    “SPARK connects what creators pay with what they produce,” said Zayven Annati, founder of Sparkvia AI. “Launching on BNB Chain extends day-one utility to a larger audience, fixed per-prompt pricing, low-fee settlement, and on-chain transparency that teams can actually operate on.”

    The $SPK sale portal is live at https://sale.sparkvia.ai/.

    What BNB Users Get Right Now

    Utility from the start: Users get Free 100 Spark credits to access 100+ writing tools, including Creative Home Page Writer, Advanced Blog Post Writer, Grammar & Style Editor, All-in-One Social Post, and more on the Sparkvia AI Writing platform.Predictable economics: Every prompt consumes the same number of credits, so budgets are clear before you click “Generate.”On-chain clarity: Top-ups and usage appear instantly in a billing dashboard and are settled on-chain, creating an auditable trail for agencies and teams.Speed to publish: Generate, refine, and export contents in minutes, top up mid-session without breaking flow.

    Since launch, Sparkvia AI has onboarded 500+ users, underscoring demand for AI tools that pair familiar workflows with crypto-native settlement.

    How to participate (BNB Chain)

    Visit https://sale.sparkvia.ai/.Follow the on-screen steps to contribute with BNB to the sale address shown.After on-chain confirmation, you will receive your $SPK token within 24 hours post sale end.

    About Sparkvia AISparkvia AI is an AI-driven writing platform delivering fast, pay-as-you-go access to advanced content tools through on-chain credits. Founded by Zayven Annati and headquartered in Malta, Sparkvia serves marketers, founders, agencies, and creators who want speed, clarity, and control in their content workflows.

    The SPARK ($SPK) presale on BNB Chain is live now at https://sale.sparkvia.ai/. For full details, participation steps, visit the $SPK sale portal.

    Sparkvia AI Socials;Website: https://sparkvia.ai/Presale Portal: https://sale.sparkvia.ai/X: https://x.com/sparkvia_AITelegram Community: https://t.me/sparkvia

    Contact details:

    Zayven AnnatiZayven@sparkvia.ai

    Disclaimer: This content is provided by Sparkvia AI. The statements, views, and opinions expressed in this content are solely those of the content provider and do not necessarily reflect the views of this media platform or its publisher. We do not endorse, verify, or guarantee the accuracy, completeness, or reliability of any information presented. We do not guarantee any claims, statements, or promises made in this article. This content is for informational purposes only and should not be considered financial, investment, or trading advice. Investing in crypto and mining-related opportunities involves significant risks, including the potential loss of capital. It is possible to lose all your capital. These products may not be suitable for everyone, and you should ensure that you understand the risks involved. Seek independent advice if necessary. Speculate only with funds that you can afford to lose. Readers are strongly encouraged to conduct their own research and consult with a qualified financial advisor before making any investment decisions. However, due to the inherently speculative nature of the blockchain sector—including cryptocurrency, NFTs, and mining—complete accuracy cannot always be guaranteed. Neither the media platform nor the publisher shall be held responsible for any fraudulent activities, misrepresentations, or financial losses arising from the content of this press release. In the event of any legal claims or charges against this article, we accept no liability or responsibility. Globenewswire does not endorse any content on this page.

    Legal Disclaimer: This media platform provides the content of this article on an “as-is” basis, without any warranties or representations of any kind, express or implied. We assume no responsibility for any inaccuracies, errors, or omissions. We do not assume any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information presented herein. Any concerns, complaints, or copyright issues related to this article should be directed to the content provider mentioned above.

    About Web3Wire Web3Wire – Information, news, press releases, events and research articles about Web3, Metaverse, Blockchain, Artificial Intelligence, Cryptocurrencies, Decentralized Finance, NFTs and Gaming. Visit Web3Wire for Web3 News and Events, Block3Wire for the latest Blockchain news and Meta3Wire to stay updated with Metaverse News.



    Source link

    When Robots Look Human, People Feel Safer—Until They Don’t – Decrypt

    By
    Metaverse Fashion Week Online
    -
    0
    When Robots Look Human, People Feel Safer—Until They Don’t – Decrypt



    In brief

    China’s Aheadform unveiled its Origin M1 robot head, reviving debate over how human is too human in machine design.
    Spanish researchers found that moderately human-like robots inspire more trust than highly realistic ones.
    Analysts project the global service-robot market will surpass $293 billion by 2032 as humanoids enter daily life.

    A new ultra-realistic robotic head has reignited the “uncanny valley” debate as humanoid machines like Tesla’s Optimus, Figure 02, and Unitree’s G1 edge closer to human form—and human discomfort.

    A Chinese robotics firm, Aheadform, unveiled a lifelike robotic head called Origin M1 that blinks, nods, and mimics facial expressions so convincingly that it unsettled viewers across social media. The clip went viral last week, racking up over 400,000 views after observers described it as “creepy” and “too real.”

    “Watching this robot head blink and follow eye movement reminded me of what Selwyn Raithe wrote in 12 Last Steps. He warned that once machines cross the line of mimicking emotion, the collapse starts quietly, not with armies, but with faces that seem more human than our neighbors,” one viewer wrote. “Chilling how close this feels.”

    That unease is what psychologists call the uncanny valley—the point where realism turns from charming to disturbing. The concept, first described by Japanese roboticist Masahiro Mori in 1970, refers to that dip in comfort as machines approach human realism without fully achieving it. The question facing designers now is how much humanity people actually want from their machines.

    With humanoid robots becoming increasingly capable and lifelike, that discomfort is rising. Tesla’s Optimus robot can now pour drinks, serve food, and perform simple factory jobs. Figure AI is pitching humanoid workers to logistics firms, showcasing its Helix robot folding laundry in a recent demonstration. At the same time, China’s Unitree G1 has drawn attention for its low cost and agile, humanlike movement.

    Scientists have begun to measure what exactly makes robots cross the invisible line between fascination and fear. A May study from Spain’s University of Castilla-La Mancha examined how human-like design influenced trust in “Bellabot,” a cat-faced delivery robot used in European restaurants.

    The researchers tested whether moderate anthropomorphism—simple facial animations and limited voice cues—made diners more comfortable with automation.

    “When robots are anthropomorphized, consumers tend to evaluate the robot more favorably,” the researchers wrote. “Anthropomorphism drives customer trust, intention to use, comfort, and enjoyment. Also, adding human attributes to a robot can make people prefer to spend more time with robots.”

    Survey data showed that empathy reduced perceived risk, but too much realism produced the opposite effect. The finding placed Bellabot in a robotics sweet spot of being friendly without being too lifelike.

    That balance is increasingly critical as humanoid robots enter commercial service. Analysts at Allied Market Research project the global service-robot market will exceed $293 billion by 2032, driven by adoption in hospitality, logistics, and healthcare.

    Across Asia and Europe, service robots like Bellabot and Bear Robotics’ Servi are already handling food delivery and table service.

    But acceptance still depends on design. A May 2025 study by Universiti Kebangsaan Malaysia, Taylor’s University, and Sunway University found that the effectiveness of service robots in restaurants depended on how well their appearance aligned with the restaurant’s service style.

    The human design problem

    Robot faces are not decorative; they’re behavioral tools. Small gestures and subtle vocal inflection can make users feel at ease, yet over-humanizing machines creates new risks.

    In 2015, a report led by MIT Media Lab ethicist Kate Darling said that people who develop empathy toward robots—especially those with names or stories—hesitated to harm them. More recently, mental health experts warned that AI and AI-powered toys designed to act human could negatively impact children’s cognitive development.

    

    “Children can form deep relationships with inanimate objects, like a teddy bear. Now you have this tool that gives you exactly what you need, because AI is going to be amazing at figuring out what you want to hear and giving that to you,” psychologist and executive coach Banu Kellner previously told Decrypt.

    That tension now defines the humanoid robot field. The Spanish researchers argued for restraint—designing robots that project just enough humanity to seem trustworthy without crossing into imitation.

    “With a robot with a high level of anthropomorphism, consumer empathy generates more negative responses and a lower level of satisfaction,” the report found. Instead, the researchers proposed designs that were less humanoid: “A service robot with a medium level of anthropomorphism positively influences the risk perceived by the consumer.”

    As humanoid robots move from spectacle to service, their success may depend less on how human they look and more on how carefully they avoid looking too human.

    Generally Intelligent Newsletter

    A weekly AI journey narrated by Gen, a generative AI model.



    Source link

    Ethereum-Based Mutuum Finance (MUTM) Records New Token Appreciation With More Than $17M Raised | Web3Wire

    By
    Metaverse Fashion Week Online
    -
    0
    Ethereum-Based Mutuum Finance (MUTM) Records New Token Appreciation With More Than M Raised | Web3Wire


    Dubai, UAE, Oct. 12, 2025 (GLOBE NEWSWIRE) — In a market where many token launches rely on brief bursts of speculative hype, Mutuum Finance (MUTM) is taking a more disciplined path. The Ethereum-based DeFi project has surpassed $17 million in funding, reflecting accelerating investor interest as it combines a clear token pricing model with visible development progress. With Phase 6 of the presale more than 60% sold out, attention is now turning to the next price step, which will coincide with the project’s upcoming testnet rollout and eventual mainnet launch.

    A Utility-Driven Protocol With Dual Lending Markets

    Mutuum Finance is a decentralized lending and borrowing protocol built to create efficient on-chain credit markets where token value is tied directly to platform usage rather than speculation. Its architecture revolves around a dual lending model that blends two complementary mechanisms.

    The Peer-to-Contract (P2C) model powers mainstream lending activity. Users can supply assets such as ETH and USDT into shared liquidity pools and receive mtTokens at a 1:1 ratio. These mtTokens automatically accrue yield generated by borrowers, functioning similarly to aTokens during Aave’s early growth. For example, a $10,000 ETH deposit could  potentially earn an average annual yield of around 15%, generating passive income while the user retains ownership of their assets.

    On the borrowing side, users can select variable or stable interest rates. Variable rates adjust dynamically with pool utilization, remaining low under moderate demand but increasing sharply as usage approaches 90% to attract new deposits and balance liquidity. Stable rates provide predictable repayment terms, though they can be rebalanced if they diverge too far from market conditions.

    Collateral management relies on clear Loan-to-Value (LTV) ratios and liquidation thresholds to maintain solvency. Stable assets like ETH and USDT typically support LTVs up to 75%, with liquidation thresholds around 80%, while more volatile tokens have stricter limits to mitigate risk. This layered approach gives both lenders and borrowers flexibility within a secure framework.

    Transparent Presale Structure and Expanding Community Tools

    Mutuum Finance’s presale model has played a central role in its steady rise. Each stage offers a fixed number of tokens at a set price. Once sold out, the next stage begins at roughly 20% higher, rewarding early buyers and creating a transparent price progression.

    The token launched at $0.01 in Phase 1 and now sits at $0.035 in Phase 6, a 250% increase for early participants. With Phase 7 set to raise the price to $0.04 and the final listing price locked at $0.06, investors have a clear view of potential upside as the sale progresses.

    To date, the presale has raised more than $17 million, allocated over 750 million tokens, and attracted 16,800+ investors. A real-time dashboard allows participants to connect wallets, track purchases, and calculate projected returns. A Top 50 leaderboard highlights major contributors, rewarding them with bonus MUTM at launch and adding a competitive edge to participation. These tools reinforce transparency while keeping engagement high as later stages approach.

    Development Roadmap and Community Initiatives

    Mutuum Finance is pairing strong fundraising momentum with clear development milestones. According to a recent team update on X, Version 1 (V1) of the protocol will launch on the Sepolia testnet in Q4 2025. This rollout will include liquidity pools, mtTokens, debt tokens, a liquidator bot, and initial support for ETH and USDT. Unlike many presales that raise capital long before delivering a product, Mutuum Finance is aligning its technical development with its fundraising timeline, building investor confidence that utility will be available shortly after listing.

    Community engagement has also been a major pillar of the rollout. A $100,000 giveaway will reward ten participants with $10,000 worth of MUTM each. Together with the leaderboard and transparency tools, these initiatives are helping build momentum as the presale enters its final stages.

    Long-Term Growth Drivers and Phase 6 Acceleration

    Looking ahead, Mutuum Finance’s roadmap features several key growth catalysts. The team plans to launch an overcollateralized stablecoin, designed to deepen platform liquidity and provide a native unit of account, an approach that helped MakerDAO scale early on. Layer-2 expansion is also planned to lower transaction fees and broaden the protocol’s reach, enabling higher throughput and lower costs for users.

    As these milestones draw closer, Phase 6 is rapidly approaching sell-out. Historically, well-structured presales with active development see demand accelerate in later stages as buyers move to secure lower entry prices ahead of listings. With 61% of Phase 6 completed and a price increase to $0.04 on the horizon, Mutuum Finance is entering a pivotal stage that will shape its trajectory into 2026.

    For more information about Mutuum Finance (MUTM) visit the links below:Website: https://www.mutuum.comLinktree: https://linktr.ee/mutuumfinance

    Disclaimer: The information provided in this press release is not a solicitation for investment, nor is it intended as investment advice, financial advice, or trading advice. Investing involves risk, including the potential loss of capital. It is strongly recommended you practice due diligence, including consultation with a professional financial advisor, before investing in or trading cryptocurrency and securities. Neither the media platform nor the publisher shall be held responsible for any fraudulent activities, misrepresentations, or financial losses arising from the content of this press release.

    About Web3Wire Web3Wire – Information, news, press releases, events and research articles about Web3, Metaverse, Blockchain, Artificial Intelligence, Cryptocurrencies, Decentralized Finance, NFTs and Gaming. Visit Web3Wire for Web3 News and Events, Block3Wire for the latest Blockchain news and Meta3Wire to stay updated with Metaverse News.



    Source link

    1...828384...224Page 83 of 224

    Popular Posts

    My Favorites

    Sony Strengthens Web3 Push with Soneium Mainnet Rollout – Cryptoflies News

    Metaverse Fashion Week Online - 0
    16Sony has rolled out the Soneium Mainnet, an Ethereum Layer 2 blockchain, as part of its growing involvement in Web3 technologies. According to a...

    Shadowheart actor begs Baldur’s Gate 3 fans not to play her...

    Donald Trump Calls Charlie Kirk a ‘Martyr for Freedom,’ Vows Justice

    The most deluxe edition of The Hobbit is $850, yet it...

    Popular Categories

    © Newsmag WordPress Theme by TagDiv