Is the Mistral API down?
Live status for the Mistral API, measured by sending a real request to mistral-small-latest every 45 seconds from our server in Chicago (US).
Uptime and time to first token come from our own tests; the official status comes from status.mistral.ai.
Live status
Mistral API at a glance
Status as of 2026-10-18 05:04 UTC: Operational.
- Uptime, 24 hours
- 100.00%
- Uptime, 7 days
- 100.00%
- Uptime, 30 days
- 100.00%
Median time to first token over the last hour: 327 ms (95th percentile 740 ms) across 80 tests of mistral-small-latest.
Last test 2026-10-18 05:04 UTC, HTTP 200.
Days with measured errors (last 90 days)
No day below 99.9% uptime since measurements began on 2026-10-18.
Official incidents (last 90 days)
None recorded from the official status page (https://status.mistral.ai).
Embed this status: badge and JSON feed. 
Check the Mistral API yourself
This sends the same one-token request we use. Set your key in the MISTRAL_API_KEY environment variable first.
curl https://api.mistral.ai/v1/chat/completions \
-H "Authorization: Bearer $MISTRAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"mistral-small-latest","max_tokens":1,"messages":[{"role":"user","content":"ping"}]}'- 200: the key and the API work.
- 401: the key is invalid or revoked.
- 429: a rate limit or exhausted quota on your account.
- 500, 502, 503: a problem on Mistral's side; check the live status above.
- Mistral’s status page has no public data feed, so independent measurements like these are the main way to see API health over time.
- The free Experiment plan is heavily rate limited; persistent 429s on your own key usually point to the plan, not an outage.
How this status is measured
Each test is a streaming chat request for a single output token. Time to first token (TTFT) runs from sending the request to receiving the first piece of generated text, so it includes network time, queueing and the model's start-up time. A request that fails with a server error or times out counts as down; a rate-limit response counts as degraded. Failures caused by our own account, such as an exhausted balance, are excluded from uptime.
Your own latency will differ with your region, model, prompt size and account tier. To watch your own endpoint and key, see monitoring and alerts.
Questions
Is the Mistral API down right now?
The card at the top of this page shows our latest test of the Mistral API, sent less than a minute ago, and the 24-hour uptime. If our tests fail and Mistral’s official status page also reports an incident, the API is having problems.
How do I check if my Mistral API key is working?
Run the curl command on this page with your key. A 200 response means the key and API work. A 401 means the key is invalid, and a 429 means a rate limit or exhausted quota on your account.
How is the Mistral API measured here?
Every 45 seconds our server in Chicago sends a real streaming request for one output token to mistral-small-latest and records whether it succeeded and how long the first token took. Uptime is the share of those requests that succeeded.
Why does this page disagree with the official Mistral status page?
Official pages usually report incidents after staff confirm them, while these tests run continuously from one location. Short or regional problems can show here first, and problems in other regions can show only on the official page.
What is a normal time to first token for mistral-small-latest?
Compare the current value with the one-hour median shown on this page. A first token several times slower than the usual median for this model is a sign of congestion even when requests still succeed.