OpenAI's Pando Git Backend Cuts Push Latency by 80%
First merged summary.
OpenAI's Pando Git backend reduces median push latency by 80% while serving nearly 1 PB of data daily for its monorepo.
OpenAI built Pando, a Rust-based Git backend for its monorepo to manage high concurrency from Codex agents. The system uses per-branch causal consistency and immutable packfiles with Express caching instead of repo-wide consistency. This achieves nearly 80 percent lower median push latency and serves over 57 TB in a single hour on standard cloud VMs. Pando maintains full Git protocol compatibility for existing workflows and enables parallel branch updates.
- Median push latency reduced by 80 percent
- Serves nearly 1 PB of Git data daily
- Per-branch causal consistency with immutable packs
- Full Git protocol compatibility retained
- Runs on commodity cloud VMs saturating 10 Gbps
Timeline
dated events in the reporting · oldest first- Nov 2, 2026OpenAI News reports Pando cuts median push latency by 5x and serves nearly 1 PB of Git data daily · OpenAI News
- Nov 3, 2026MarkTechPost reports Pando achieves 80 percent lower push latency using causal consistency · MarkTechPost
What each outlet adds
beyond the shared facts- OpenAI NewsHighlights parallel branch updates and full Git client compatibility.
- MarkTechPostDetails on per-branch causal consistency, immutable packs, 57 TB per hour throughput, and cloud VM performance.
Coverage
oldest first · one row per article · UTC- · 18h ago
OpenAI NewsFirst report
OpenAI builds Pando for scaling Git monorepo
OpenAI's Pando backend cuts median push latency by 5x while serving petabytes of Git data daily for Codex-driven development.
- · 3h ago
MarkTechPost
OpenAI Introduces Pando Git Backend Cutting Push Latency 80%
Pando serves nearly 1 PB of Git data daily for OpenAI's monorepo with causal consistency and pack caching.