# Riemann Computing Releases 13B Parameter AI Model Pre-release and Other News | Riemann Computing

- Canonical URL: https://wefunder.com/feed/430336
- Entity ID: wefunder:feed_item:430336
- Published at: 2026-10-21 00:32:57 UTC
- Updated at: 2026-10-21 00:33:00 UTC

## Author
Andrew Magdy Kamal

## Subject
Riemann Computing

## Content
Riemann Computing is excited to announce the prerelease of "My Name is Riemann" via Huggingface and its demo space along with the demo spaces for NTK Trainer and OpenPeerLLM. My Name is Riemann represents a 13.1 billion parameter decoder-only language model that has been progressively developed through successive iterations from the publicly available weights of OpenPeerAI/OpenPeerLLM and subsequently refined utilizing the OpenPeer-AI-NTK-Trainer gate controller mechanism in lieu of LoRA methodology. This constitutes the inaugural public release within the My Name is Riemann series and has been disseminated in prerelease status. We released this utilizing less than $3K worth of hardware for the first training stage locally, and about $640 worth of GPU compute on the cloud. The metrics displayed above might prompt you to ask, "What makes this valuable?" or "Can you break this down for someone without technical expertise?"Here's the straightforward answer: Riemann Computing has developed and successfully demonstrated a more economical approach to training and computational efficiency. Despite working with a constrained dataset, limited iterations, entropy considerations, and limited hardening, the level of parameters achieved at this price point is remarkably uncommon within the AI sector. What's particularly impressive is that we accomplished these results without utilizing our proprietary compression, LonScript, signal computing, or our advanced AI capabilities. Instead, we relied solely on Neural Tangent Kernel (NTK) methodology combined with less conventional statistical approaches designed for budget-conscious training and computation.This demonstrates that Riemann Computing possesses substantial competitive advantage in transforming the AI landscape. We're positioned to deliver companies a comprehensive platform encompassing cost-effective training, efficient data storage, optimized computing infrastructure, and comprehensive telemetry. This is now starting to be all grounded in proven, practical real-world applications rather than theoretical concepts alone.As always, none of this is investment or financial advice. Please do your own due diligence and research.