OpenAI details Pando, its Git backend, and a near-80% cut in median push latency
What happened
On November 2, OpenAI published the architecture and deployment results for Pando, its Git backend. The team used Codex to go from project start to production deployment in 10 weeks. Pando now supports all development in OpenAI's monorepo, and median push latency in production has dropped by nearly 80%. The system stays compatible with the Git protocol and clients. It scales reads and writes by coordinating updates per branch, reusing packfiles and caching by region, which supports agent-scale parallel development. It does not give a consistent snapshot of all branches at one moment. OpenAI also contributed some Git client improvements upstream.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- OpenAI NewsScaling Git at OpenAI
OpenAI 详解 Git 后端 Pando 的架构，该系统已支撑其单体仓库的全部开发，生产环境中推送延迟中位数降低近 80%。Pando 保持 Git 协议和客户端兼容，通过按分支协调更新、复用 packfile 和区域缓存扩展读写能力，但不提供所有分支在同一时刻的一致快照。团队借助 Codex 在 10 周内完成从启动到生产部署，并将部分 Git 客户端改进贡献给上游。
Heat over time
Heat now 7·Comparable peak 8（Nov 3 08:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.