Mistral AI
European AI lab founded by ex-DeepMind and ex-Meta researchers. Builds open-weight and commercial models with a focus on efficient inference and on-prem deployability. Distributed primarily through La Plateforme (its API) and Le Chat (its consumer assistant).
Products
Le Chat · La Plateforme · Mistral / Mixtral / Codestral / Magistral models · Le Console
Reliability — last 30d
Composite score
Uptime
Incidents
Median MTTR
What's breaking (last 30d)
Components ranked by incident count. Click a row for the incident list. Down-time is attributed proportionally for multi-component incidents.
| Component | Incidents | Down (attr.) | Median MTTR | Last |
|---|---|---|---|---|
Chat Completions API | 94 | 9h 4m | 1m | 4h ago |
Audio API | 39 | 3h 7m | 1m | 7d ago |
Embeddings API | 21 | 40m | 1m | 3d ago |
Conversations API | 13 | 1h 22m | 5m | 4h ago |
OCR API | 9 | 26m | 1m | 6d ago |
Files API | 8 | 2h 5m | 10m | 18d ago |
Batch API | 5 | 1h 1m | 5m | 1d ago |
Le Chat | 4 | 58m | 7m | 20d ago |
Score breakdown
SLA compliance
How often does this provider clear their stated availability target?
Incident history
showing 27 of 200 · 4 of 29 days
Type of failure
Categorised from the public incident headlines (outage / latency / auth / etc.). Click a category to drill in.
When it breaks
Incident starts by weekday × hour, UTC. Hover a cell for incident detail.
Hour of day · UTC
Most likely outage time
Hottest hour
Worst day
Worst time of day
How fast they fix things
Distribution of incident resolution times.
Typical fix
2m
half of incidents resolve faster
On a bad day
13m
9 in 10 resolve faster than this
Distribution
+1 resolved w/o published end time (excluded)
Mistral AI — incidents by day (last 30d)
Worst incidents
Ranked by severity-weighted duration (impact × wall-clock). Click any incident for the full update timeline.
Method: a rolling quarter (13 weeks ≈ 91 days). Each cell is one 7-day slice of Statuspage-standard uptime (1 − (critical + 0.3 × major) / window; minors and maintenance excluded); red when the slice dips below the stated SLA.
* No public SLA — using common enterprise target (99.9%) for comparison.