OpenAI is investigating an “elevated latency” issue affecting both ChatGPT and its API, meaning some users may see slower responses rather than a clean service failure.
The issue was reported Wednesday morning, with OpenAI’s system status shown as of 8:15 am PT, according to 9to5Mac . A separate issue involving FedRAMP users was also being monitored, though 9to5Mac reported that a fix had already been applied for that incident.
“Team is investigating elevated latencies across the board for API and ChatGPT. Updates will be posted on https://t.co/R3dCKGH1G4”
— Tibo (@thsottiaux), May 27, 2026
OpenAI investigates elevated latency slowing ChatGPT responses
The key phrase is “elevated latency.” That points to degraded performance — slower replies, longer waits, delayed completions — not necessarily a total outage.
For users, the symptom is simple: ChatGPT may feel unusually slow today. The cause may not be a local Wi-Fi problem, browser issue, or device slowdown. Based on OpenAI’s current description, the company is looking into a service-side performance problem affecting ChatGPT and the API.
The scope is still not fully detailed. OpenAI has not yet confirmed a root cause, a restoration timeline, or whether the issue is concentrated in any specific region, plan, model, or traffic pattern.
That distinction matters. A latency incident can leave the product technically available while making it painful to use. For interactive ChatGPT sessions, that means waiting longer for an answer. For API customers, it can mean slower application responses and a higher chance that dependent systems bump into their own timeout limits.
Here is the clean read from the available status information:
| Service area | Reported condition | Confirmed fix? |
|---|---|---|
| ChatGPT | Elevated latency under investigation | Not yet reported in source |
| OpenAI API | Elevated latency under investigation | Not yet reported in source |
| FedRAMP users | Separate issue monitored | Fix applied, per 9to5Mac |
The FedRAMP issue should not be conflated with the broader ChatGPT and API latency report. The source describes them as separate items.
ChatGPT slowdown could disrupt work, coding and customer support workflows
For people using ChatGPT casually, latency is annoying. For developers and businesses building around the OpenAI API, it can be operationally awkward.
A slow model response can delay writing drafts, coding help, research tasks, brainstorming sessions, data analysis prompts, or customer-facing flows that depend on fast completions. The source does not say those workflows are broken. The risk is that degraded response time can ripple through anything waiting on ChatGPT to finish.
MLXIO analysis: the API side is the more sensitive part of this report. A consumer can wait a few extra seconds and retry. An application wired into an API may have stricter expectations around response timing, queue depth, and retry behavior. If latency stretches far enough, downstream software can interpret slowness as failure even when OpenAI is still returning valid responses.
That said, there is no confirmed evidence yet of a full outage. There is also no published severity level in the provided source beyond the language that OpenAI is investigating “elevated latencies across the board for API and ChatGPT.”
This is not the same kind of OpenAI story as the company’s legal fights, including our earlier coverage of Two Hours Crushed Musk’s OpenAI Case Against Altman and $38M Fight Dies as Musk's OpenAI Lawsuit Runs Late. Today’s issue is narrower and more immediate: whether the service can respond quickly enough for users and developers who depend on it right now.
A recent incident gives some limited context. An IsDown incident report cited an OpenAI issue on May 21, 2026 involving “elevated latency and error rates” for ChatGPT 5.5 Thinking, which it said lasted 1.4 hours and was resolved after mitigation. That does not explain today’s issue, but it shows the kind of status trail engineers and customers will look for: identification, mitigation, monitoring, then resolution.
OpenAI status updates will signal whether ChatGPT latency is easing
The next meaningful update will be whether OpenAI identifies the cause, deploys a fix, or moves the incident into monitoring. Until then, the practical read is limited: ChatGPT and API performance may be degraded, and OpenAI is investigating.
Latency incidents can stem from several places, including traffic spikes, backend service issues, model-serving constraints, or infrastructure changes. No specific cause has been confirmed here, so any sharper explanation would be speculation.
For users trying to work through the slowdown:
- Retry later: If a prompt hangs, waiting may be better than repeatedly resubmitting.
- Refresh carefully: Reloading a session may help rule out a local browser issue, but it will not fix provider-side latency.
- Check basics: Confirm your own connection before assuming every delay is on OpenAI’s side.
- Avoid prompt spam: Re-sending long prompts over and over can create duplicated work and confusion if delayed responses eventually arrive.
For developers, the useful signal is not just whether ChatGPT loads. It is whether API latency returns to a range their own applications can tolerate. Watch internal telemetry alongside OpenAI’s public status updates.
MLXIO will update this story if OpenAI confirms a root cause, publishes a restoration timeline, or expands the list of affected services. The immediate watch item is simple: whether the incident moves from investigation to mitigation — and how long users remain stuck waiting for ChatGPT to answer.
The Bottom Line
- ChatGPT and API users may experience slower responses even if the services remain available.
- Developers relying on OpenAI’s API could see app delays or timeout issues.
- OpenAI has not yet confirmed the root cause, affected regions, or restoration timeline.
