# Is an AI service down? How to check the evidence

Compare official AI status, collection freshness and your failed request. Learn why a normal provider report cannot confirm every model or network.

By IsAnythingDown. Reviewed 2026-10-05.

Canonical page: https://isanythingdown.com/guides/is-ai-service-down

## Quick answer

Start with the exact product and failed operation. Compare its official report with your request error and the report’s collection time. A reported incident is useful evidence; a normal report alone cannot establish that your model, account or network works.

## 1. Check the product you actually used

ChatGPT and the OpenAI API have separate entries here, derived from their respective official components. A ChatGPT sign-in problem is not evidence that an API request failed. A provider-wide headline also cannot identify the result for a particular model or account.

Gemini app reports come from Google Workspace’s Gemini incident feed. Gemini API / AI Studio has a separate official status page, linked here without automatic collection. For OpenRouter, check the router’s page as well as the selected provider when that provider is known; the two reports describe different services.

- [IsAnythingDown sources & methodology](https://isanythingdown.com/methodology)
- [Gemini API / AI Studio official status](https://aistudio.google.com/status)
- [OpenRouter official status](https://status.openrouter.ai/)

## 2. Read both timestamps

Collected at tells you when IsAnythingDown last received and validated the report. Source updated at comes from the provider. An unchanged source timestamp can coexist with successful recent collections, so it is not itself evidence of a collection failure.

Official reports become stale here after six minutes without a valid collection. Stale data describes the last saved report. A failed collection is shown separately and is not converted into a provider outage. If the database cannot be read, the page says Data unavailable.

- [IsAnythingDown sources & methodology](https://isanythingdown.com/methodology)

## 3. Compare the report with the failed request

Record the UTC time, endpoint, model, HTTP status, provider error code and request ID where available. A 401, a quota-related 429 and a server error call for different checks. Use the provider’s error documentation rather than treating every failure as downtime.

If an incident names the component you used and overlaps your request time, it may explain the failure. That remains a possible association unless the provider or additional evidence establishes the cause. If there is no relevant incident, continue checking the request and its network path.

- [OpenAI API error codes](https://developers.openai.com/api/docs/guides/error-codes)
- [Claude API errors](https://platform.claude.com/docs/en/api/errors)
- [Gemini API troubleshooting](https://ai.google.dev/gemini-api/docs/troubleshooting)

## 4. Check what was independently observed

IsAnythingDown configures independent checks for the GitHub public homepage and one public repository metadata endpoint. The center does not run these checks. Previous deployment-server checks are stopped and do not establish regional coverage. No remote regional nodes or AI inference probes are deployed.

A status-page fetch, a reachable product homepage or another user’s successful request does not reproduce your own API operation. If you run a diagnostic request in your own environment, keep its target, time, resolver and network visible, and consider request cost and side effects before repeating it.

- [Actual probe coverage](https://isanythingdown.com/coverage)

## What does “no reported incidents” mean?

For Gemini app, it means no active relevant incident appeared in the collected feed. It is not an independent confirmation of availability. More generally, an empty event list only describes the saved report. A report that does not supply event lists is labelled as not provided, rather than as incident-free.

- [IsAnythingDown sources & methodology](https://isanythingdown.com/methodology)

## Related service pages

- [Claude](https://isanythingdown.com/services/claude)
- [ChatGPT](https://isanythingdown.com/services/chatgpt)
- [OpenAI API](https://isanythingdown.com/services/openai)
- [Gemini](https://isanythingdown.com/services/gemini)
- [Hugging Face](https://isanythingdown.com/services/hugging-face)
- [OpenRouter](https://isanythingdown.com/services/openrouter)

Each entry shows its own collection coverage. A linked service is not necessarily automatically collected.

Editorial review date applies to this guide. Live report collection times appear on the service pages.