Downrightnow
// Incident Post-MortemDRN-80E29609

Replicate outage on June 3, 2026

Major severityResolved2h 58mAI Tools
Started · UTC
Jun 03, 05:59 PM
4h ago
Resolved · UTC
Jun 03, 08:57 PM
1h ago
Duration
2h 58m
End-to-end
Peak confidence
80.00
75 snapshots
// User impact

Replicate users likely saw failed connections, degraded availability, or session instability for this AI Tools service. The incident reached Major severity and lasted 2h 58m. The stored report window does not show a preserved user-report spike.

// What happened

Replicate appears to be operating normally based on current signals. DNS is no recent DNS check, user reports are quiet.

Cached · Wed, Jun 03, 2026, 09:00 PM UTC · scope: this incident

// Timeline

  1. Incident detected

    Major confidence crossed the incident threshold for Replicate: we're seeing long setup times and high contention for models on some L40S and H200 clusters..

  2. Status changed to Partial Outage

    The stored readout moved to Partial Outage with confidence 38.00.

  3. Status changed to Possible Regional Issues

    The stored readout moved to Possible Regional Issues with confidence 38.00.

  4. Status changed to Partial Outage

    The stored readout moved to Partial Outage with confidence 38.00.

  5. Marked resolved

    The incident was closed after 1h 14m of observed impact.

  6. Peak confidence reached

    Confidence peaked at 80.00 while the service was marked Operational.

// Signals by source0 sources

No per-source signal payloads were preserved in this incident's snapshot window.

// Aggregate evidence

Confidence scoring

80.00

The incident reached 80.00 peak confidence across 75 stored status snapshots.

User reports

0 / 15m

No elevated user-report volume was preserved in the incident window snapshots.

Status samples

0 affected

The incident row exists, but the bounded snapshot window does not include affected status samples.

Recovery signal

Resolved

Recovery was recorded at Wed, Jun 03, 2026, 08:57 PM UTC.