yarn

Powerful AI shouldn't require everyone to own expensive hardware. Everyone wants to run advanced AI models locally. But there's a reality that marketing often overlooks. Running an AI model isn't just about downloading it. You need to understand model size, quantization (Q4, Q5, Q8 or FP16), memory requirements, and actual inference performance measured in tokens per second. A model running locally doesn't necessarily mean a fast or usable experience. There's another problem we rarely talk about. Companies and individuals invest thousands of dollars in professional AI hardware, such as NVIDIA RTX PRO 6000 GPUs. Yet these expensive resources can sit idle for hours or even days when they're not being used. Meanwhile, someone else is saving money to purchase exactly the same hardware, only to have it sit idle when their own workloads are finished. We're duplicating expensive infrastructure instead of using what already exists. This is what we're building with MineBench IQ. A distributed AI compute marketplace that connects available computing resources with people who need them. Have a data center or powerful multi-GPU infrastructure? Become a provider. Host large models such as DeepSeek, Kimi or Qwen, share your inference capacity, offer different performance tiers measured in tokens per second, and earn from resources that would otherwise remain unused. Have a powerful gaming GPU or workstation? You can contribute too! Run smaller language models, coding assistants, image-generation models such as FLUX, and other AI workloads. You don't need enterprise infrastructure to become part of the network. What if you need access to powerful AI? You don't need to buy expensive hardware yourself. Access models and computing resources from providers across the network. Need Qwen for coding? Access a provider running it. Need DeepSeek for reasoning? Use infrastructure capable of hosting it. Want to generate images with FLUX? Connect to a provider with suitable resources. Need more memory or higher inference throughput than your local machine can provide? Access larger computing infrastructure. Our vision is to make better use of the hardware we already have. Providers monetize idle computing capacity. Users get access to powerful AI without purchasing expensive equipment. Developers get access to a distributed pool of computing resources. Instead of everyone buying the same expensive hardware and leaving it unused, we can build a shared ecosystem where computing power is available when and where it's needed. MineBench IQ - turning idle computing power into efficient computing power.