Is the Mistral API reachable right now?
A live probe of api.mistral.ai from NetOkay’s edge, a test from your own browser, and the checks that separate a Mistral incident from a problem between you and the EU.
In short: The card below shows the latest probe of api.mistral.ai from
eight Cloudflare regions: one GET for /v1/models with a placeholder token and
expects the API’s own 401. Then test from this browser: a long round trip to Europe makes
idle and total timeouts on your path bite sooner than they do for US‑hosted providers.
Last updated
Live probe · api.mistral.ai
Where is api.mistral.ai reachable from?
Loading the latest observation of api.mistral.ai… Test from this browser →
From this browser
—Not run. One request from your browser, sent only when you ask.
What this proves, and what it does not
Loading…
Mistral specifics
- API host
-
api.mistral.ai, HTTPS on port 443, with an OpenAI‑style/v1/surface. Mistral’s platform is hosted in the EU, so clients far from Europe see higher round‑trip times than to US‑hosted providers. - Streaming
-
Server-sent events with
data:JSON chunks and a finaldata: [DONE], compatible with OpenAI clients. Setstream: true; without it the whole completion arrives in one JSON body. - Latency and timeouts
- A 150–250 ms round trip from Asia or the Americas is normal. Client libraries with short connect or read timeouts, and proxies with aggressive idle limits, fail here first even though the API is healthy.
- Official status
- status.mistral.ai publishes platform incidents.
- Client defaults
-
The official Python client hard-codes a 5-minute timeout on chat calls and does not
retry unless you pass a
RetryConfig; the TypeScript client leaves retries to the API. With no automatic retry, a single 429 or 5xx surfaces directly in your application.
What Mistral status codes mean
| Code | Sent by | Meaning |
|---|---|---|
| 401 | API application | Missing or invalid key. The API application produced this answer. |
| 422 | API application | The request body failed validation, a frequent surprise when switching from another provider’s parameters. |
| 429 | API application | Rate or spend limit reached. |
| 5xx | API application | Server-side error. Retry with backoff. |
| Every row is an HTTP answer: the host was reached and the request was refused or served. None of them means api.mistral.ai is unreachable. | ||
Reachable here but failing for you?
- Run the 30‑second idle stream from the network where your app fails. A long completion from an EU host is exactly the kind of quiet connection a proxy idle limit cuts. Run the idle stream →
- Raise client timeouts before blaming the API: a 250 ms round trip plus a slow first token can exceed a default connect or read timeout. The streaming guide covers idle limits and buffering. Fixes in the streaming guide →
- Compare environments if this browser passes but a container, server or agent fails, then run the CLI from the failing environment against api.mistral.ai. Compare environments →
Mistral API status FAQ
My requests time out only for Mistral. Why?
Distance. If your client sets one timeout for every provider, the EU round trip plus model latency can exceed it. Compare the probe’s response time with your client’s limit before assuming an outage.
What is a 422 from Mistral?
A validation error: an unsupported parameter, a bad message role or an unknown model name. It is an HTTP response, so connectivity is fine; fix the request body.
Is this an uptime monitor?
No. One recent observation per minute per Cloudflare location, no history. Use status.mistral.ai for incident reports.