OpenAI has introduced Pando, a new Git backend that now serves all development in its monorepo. The OpenAI Pando Git backend preserves the standard Git client and protocol: engineers and Codex agents still run git push and git fetch exactly as before. Only the server changed. According to OpenAI, median push latency has fallen by nearly 80%, and Pando now delivers close to a petabyte of Git data per day.
TL;DR
- Scale: Nearly 1 PB of Git data per day, with peaks above 57 TB in a single hour.
- Runs on: Commodity hardware; standard cloud VMs saturate 10 Gbps links on clone and fetch operations.
- Performance: Supports thousands of concurrent pushers without degradation, based on OpenAI's own testing.
- Best result: Median push latency down nearly 80% (roughly 5x faster).
- Limitations: Not open source; repeated branch pushes resend non-trunk history (p99 pack size remains under 3 MiB).
What Is OpenAI Pando?
Pando is a Git-compatible server backend from OpenAI designed to scale concurrent pushes and fetches for a large monorepo. It is written in Rust and reached production in just 10 weeks. One engineer designed the core architecture, while teams of Codex agents implemented it. Pando has served production traffic since April 2026.
Why It Matters in 2026
As AI coding agents such as Codex increasingly operate directly on source repositories, the volume and concurrency of Git operations have grown far beyond what traditional Git hosting was designed to handle. Pando reflects a broader 2026 trend: AI infrastructure teams are re-engineering developer tooling itself to keep pace with agentic workflows, where hundreds or thousands of automated contributors may push and fetch simultaneously. By keeping the client-facing Git protocol unchanged, OpenAI avoids disrupting existing developer habits while modernizing the backend for petabyte-scale, high-concurrency workloads on commodity cloud hardware.
via MarkTechPost