Is ChatGPT down? Not right now.

OpenAI's status page says "All Systems Operational" (read Oct 29, 10:30 UTC). Below: ChatGPT, the OpenAI API and Codex one by one, every recent incident, how each GPT model is answering on Zurelay, and what to do while ChatGPT is down.

OpenAI's status

All Systems Operational
  • APIs14Operational
    • Chat CompletionsOperational
    • ResponsesOperational
    • AgentsOperational
    • DecisionsOperational
    • Fine-tuningOperational
    • EmbeddingsOperational
    • ImagesOperational
    • BatchOperational
    • AudioOperational
    • ModerationsOperational
    • RealtimeOperational
    • FilesOperational
    • LoginOperational
    • LiveOperational
  • ChatGPT16Operational
    • ConversationsOperational
    • LoginOperational
    • ChatGPT WorkOperational
    • Codex in ChatGPT DesktopOperational
    • Compliance APIOperational
    • SearchOperational
    • File uploadsOperational
    • Voice modeOperational
    • GPTsOperational
    • Image GenerationOperational
    • Deep ResearchOperational
    • AgentOperational
    • dotsOperational
    • SpaceOperational
    • SitesOperational
    • Connectors/AppsOperational
  • Codex4Operational
    • Codex WebOperational
    • Codex APIOperational
    • CLIOperational
    • VS Code extensionOperational
  • FedRAMP1Operational
    • FedRAMPOperational
  • Ads Platform2Operational
    • Ads ManagerOperational
    • Ads APIOperational

From status.openai.com, read Oct 29, 10:30 UTC and again every minute while this page is open.

GPT on Zurelay

All answering
  • GPT-6 Astra99.93% answered in 7 days · first token 4.7 s
    Operational
  • GPT-6.1 Sol100% answered in 7 days · first token 4.0 s
    Operational
  • GPT-6.1 Terra100% answered in 7 days · first token 4.5 s
    Operational
  • GPT-6 Sol100% answered in 7 days · first token 3.7 s
    Operational
  • GPT-6 Luna99.92% answered in 7 days · first token 3.7 s
    Operational
  • GPT-5.6 Sol100% answered in 7 days · first token 3.9 s
    Operational
  • GPT-5.6 Terra100% answered in 7 days · first token 4.2 s
    Operational
  • GPT-5.6 Luna94.2% answered in 7 days · first token 2.2 s
    Operational

Zurelay's own routes for each model, checked every few minutes and measured on real traffic, separately from OpenAI's page. Every model's 90-day history is on the Zurelay status page.

History

Recent OpenAI incidents. As OpenAI posted them.

Since October 7, OpenAI posted 25 incidents across ChatGPT, the API, Codex and its other products, 161 h 43 min in all. Half were over within 1 h 34 min, and 4 were rated major or worse; most affected one feature or model.

What to do

When ChatGPT is down. Or just slow.

In ChatGPT, open the ChatGPT group above. If conversations or login show an incident, there's nothing to fix on your side; the incident's page has OpenAI's updates. If only one feature is listed, such as voice mode, image generation or connectors, the rest of ChatGPT usually keeps working.

On the API, the error tells you whose problem it is:

StatusTypeWhat it means
503server_is_overloadedThe model is overloaded. Follow Retry-After, then retry.
500server errorAn error on OpenAI's servers. Retry after a short wait.
429rate limit reachedYou're sending requests too fast. Follow Retry-After.
429slow_downYour request rate rose too quickly. Ramp up gradually.
429credit_balance_exhaustedNo prepaid credit left. Add credit; retrying won't help.
401incorrect API keyThe key is wrong or revoked, not an outage.

Retry 5xx errors after a short wait, following Retry-After when it's there, and use exponential backoff with jitter when it isn't. Some incidents name a single model, so falling back to another GPT model keeps most apps running.

On Zurelay, each GPT model is served over several routes run by independent sellers, and a request that fails on one moves to the next; attempts that fail aren't billed. With smart routing on, the closest other model answers if every route for a model is down. It's the same Chat Completions API, so the OpenAI SDKs work with a new base URL; see GPT API pricing.

FAQ

ChatGPT status questions.

Anything else? Ask our team and we’ll answer by email.

Is ChatGPT down right now?

When this page was built, at Oct 29, 10:30 UTC, OpenAI's status page said "All Systems Operational". Open this page to see ChatGPT, the API and Codex one by one; it checks OpenAI's status again every minute while it's open.

Is the OpenAI API down, or just ChatGPT?

OpenAI tracks them separately: the APIs (Chat Completions, Responses, Realtime, Images and more), ChatGPT (conversations, login, voice mode, image generation, connectors and more) and Codex each have their own status. A ChatGPT incident can leave the API working, and the other way round. The table at the top of this page shows each group.

How often does ChatGPT go down?

Since October 7, OpenAI posted 25 incidents across all its products, 161 h 43 min in all; half were over within 1 h 34 min, and 4 were rated major or worse. Most affected one feature or model rather than all of ChatGPT.

What does OpenAI's 503 server_is_overloaded error mean?

It means the model you asked for is temporarily overloaded, not that your request or key is wrong. Follow the Retry-After header when there is one, then retry; if it keeps happening, check OpenAI's status page for an incident.

Why do I get 429 errors when OpenAI isn't down?

A 429 is about your own account: requests sent too fast, a request rate that rose too quickly (slow_down), no prepaid credit left (credit_balance_exhausted), or a spend or usage limit. Rate limits pass with time; OpenAI notes that retrying billing, spend or quota errors won't restore access.

Where is OpenAI's official status page?

At https://status.openai.com. This page reads the same data and adds how each GPT model is answering on Zurelay's own routes.

More pricing

Stop paying list price.
Start saving today.

Create a key in seconds, point the SDK you already use at Zurelay, and every request costs up to 90% less from the first token.

Get your API keyTalk to an engineer