Field notes · October 2026

How the proof was built.

Practical field notes from building, breaking, operating, and evolving a self-hosted open-model system.

Article 1October 2026

How We Ran an Open-Source LLM on a Cloud GPU

RunPod + Ollama + Qwen3 4B: a practical proof, not a benchmark

Read article →
Article 2October 2026

When the Demo Worked - and Then Everything Broke

What persistent storage, cloud networking, and GPU detection taught us

Read article →
Article 3October 2026

Making a Self-Hosted LLM Easy to Restart Was Harder Than Running It

What RunPod storage, GPU availability, persistent models, HTTP proxying, and cold starts taught us about practical operation

Read article →
Article 4October 2026

Why Does Running an Open Model Cost So Much?

What we learned about GPU economics, CPU hosting, hosted inference, and why RunPod was a reasonable first choice for Pegas — even though it was not the final infrastructure answer.

Read article →
Article 5October 2026

The Day Pegas Stopped Needing a GPU Babysitter

How request-driven serverless inference, scale-to-zero, and a stable Modal endpoint changed the operating model of Pegas.

Read article →
Article 6October 2026

From One Model Call to a Multi-Agent System: How Pegas Learned to Route Work

How Pegas became a visible Request Desk with specialized agents, conditional privacy review, bounded correction loops, and inspectable handoffs.

Read article →
Article 7October 2026

From AI Workflow to Agentic Finance: How Pegas Learned to Pay for Expertise

How bounded authority, x402 on Solana Devnet, Alchemy evidence, Upstash payment state, and ideas from Lead's agentic finance framework became one working Pegas flow.

Read article →
Article 8 · LatestOctober 2026

Pegas Became an MCP Server: Other Agents Get a Tool, Not Authority

How Pegas opened itself to external AI agents through the Model Context Protocol, and why the caller gets a tool but never payment authority.

Read article →