OpenAI is wrestling with a partial outage that's left users unable to access ChatGPT, Codex, and the API Platform. The incident started around 2:35 PM on April 20, 2026, with 12 ChatGPT components and one Codex component showing issues. As of 3:29 PM, OpenAI's team was still investigating the root cause, according to the company's status page.
But this might not be just an OpenAI problem. Users on Hacker News reported Reddit having loading issues around the same time, and Downdetector showed spikes across multiple services. One commenter pointed to possible DNS issues as the common thread. If that's accurate, the OpenAI outage could be part of a larger internet infrastructure failure rather than an isolated incident.
When ChatGPT goes dark, users scatter to Claude or Gemini. But developers with production systems tied to OpenAI's API are stuck waiting. The switching cost is real: new API endpoints and different prompt formats can break production systems. Some companies have started building fallback routes to other providers, but that takes engineering work most teams haven't prioritized.
OpenAI's status page shows its services typically run at 99.85 to 99.99% uptime. What's unusual about this incident is that it hit multiple core products simultaneously. The status page notes availability metrics are reported at an aggregate level, so actual impact varies by subscription tier and usage.
If the DNS theory checks out, OpenAI is collateral damage in a bigger infrastructure mess.