A 30B model on a Raspberry Pi

PLUS: A new AI coding framework and the developer revolt against Copilot

Author

Gen AI Art
September 07, 2025 • Estimated Reading Time: 9 minutes

Together with

The notion of running large-scale AI models has long been tied to expensive, enterprise-grade hardware. A new open-source project, however, is challenging that convention by successfully running a 30-billion-parameter model on a small cluster of four Raspberry Pi 5s.

This experiment achieved practical speeds on affordable, off-the-shelf components, proving the system is performant enough for real-world tasks. Does this breakthrough signal a new era of decentralized AI, moving powerful capabilities out of the data center and into the hands of developers and hobbyists?

Today in AI:

  • Running a 30B model on a Raspberry Pi cluster

  • A new open-source AI coding framework

  • The developer revolt against GitHub Copilot

PRESENTED BY THE AI REPORT

AI You’ll Actually Understand

Cut through the noise. The AI Report makes AI clear, practical, and useful—without needing a technical background.

Join 400,000+ professionals mastering AI in minutes a day.

Stay informed. Stay ahead.

No fluff—just results.

What’s new? An open-source project successfully ran a 30-billion-parameter AI model on a cluster of four Raspberry Pi 5s. This experiment showcases how distributed computing can bring large model inference to affordable, off-the-shelf hardware.

What matters?

  • The entire setup runs on just four Raspberry Pi 5 devices, each with 8GB of RAM, connected by a simple network switch.

  • This tiny cluster achieved a practical prediction speed of 13.04 tokens per second, demonstrating that the system is performant enough for real-world tasks.

  • The achievement was made possible by distributed-llama, an open-source framework designed to intelligently split a model's workload across multiple smaller computers.Why it matters?

This experiment proves that running powerful AI models is no longer limited to expensive, enterprise-grade hardware. It opens up new possibilities for developers and hobbyists to build complex AI applications at the edge using accessible components.

What’s new? A new open-source methodology, Disciplined AI Software Development, offers a structured framework for developers to collaborate with AI, aiming to prevent common issues like code bloat and architectural drift.

What matters?

  • The framework guides developers through four distinct stages—AI configuration, collaborative planning, systematic implementation, and iteration—to treat the AI as a structured collaborator rather than just a code generator.

  • A core principle is a strict rule where each file stays ≤150 lines, which keeps context windows small for the AI and makes code easier for developers to validate and debug.

  • Instead of guesswork, the methodology pushes for data-driven iteration by requiring developers to build a benchmarking suite first, allowing them to measure performance and feed real data back to the AI for optimization.

Why it matters?

This approach shifts the developer's role from simply prompting an AI to actively architecting a collaborative workflow. Adopting this discipline can lead to more maintainable and reliable software as AI becomes a central part of the development process.

PRESENTED BY 1440 MEDIA

Looking for unbiased, fact-based news? Join 1440 today.

Join over 4 million Americans who start their day with 1440 – your daily digest for unbiased, fact-centric news. From politics to sports, we cover it all by analyzing over 100 sources. Our concise, 5-minute read lands in your inbox each morning at no cost. Experience news without the noise; let 1440 help you make up your own mind. Sign up now and invite your friends and family to be part of the informed.

What’s new? A growing contingent of the developer community is pushing back against GitHub’s forced integration of Copilot, with vocal complaints and calls for opt-out measures dominating platform discussions.

What matters?

  • The two most popular GitHub community discussions over the past year are threads from developers asking for ways to block AI-generated pull requests and disable unwanted Copilot features.

  • This frustration is translating into action, with developers actively migrating projects to open-source alternatives like Codeberg, a move long encouraged by the Software Freedom Conservancy’s Give Up GitHub campaign.

  • While Microsoft reports strong momentum with 20 million Copilot users, many developers feel the company is prioritizing AI metrics over user consent and the integrity of their open-source repositories.

Why it matters?

This clash highlights a fundamental tension between a platform’s aggressive AI strategy and its community’s desire for control. If the developer exodus continues, it could begin to chip away at the powerful network effects that have secured GitHub’s dominance.

Everything else in AI

OpenAI published a new research paper arguing that hallucinations persist because current training and evaluation methods incentivize models to guess rather than admit uncertainty.

DeepSeek targets an end-of-year release for its new AI agent, which is designed to learn from experience and rival offerings from OpenAI.

Apple lost its lead AI researcher for robotics to Meta's Robotics Studio, the latest in a series of talent departures from Apple's AI teams this year.

NewsGuard found in a new study that chatbot misinformation rates have jumped from 18% to 35% in 2025, raising alarms about the reliability of AI models during breaking news events.

PRESENTED BY I HATE IT HERE

The free newsletter making HR less lonely

The best HR advice comes from those in the trenches. That’s what this is: real-world HR insights delivered in a newsletter from Hebba Youssef, a Chief People Officer who’s been there. Practical, real strategies with a dash of humor. Because HR shouldn’t be thankless—and you shouldn’t be alone in it.

Essential AI Guides - Reading List:

Your opinion matters!

Work with us

Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now →