About

BrokenAI

A live head-to-head reliability scoreboard for the major AI providers. Numbers are pulled straight from each provider's official Statuspage feed every minute — we don't run synthetic probes, so what you see is what they reported.

See a number that looks wrong, or want a provider added? Ping me on @threatner_verified.

Project

  • About & scoring formula
  • 30-day report
  • SLA compliance
  • Embed badges

Providers

  • Claude

    Reliability profile →Official Statuspage

  • OpenAI

    Reliability profile →Official Statuspage
Not affiliated with Anthropic or OpenAI. All trademarks belong to their respective owners.@threatner_verified·© 2026 BrokenAI
BrokenAISee 13-week SLA →

30-day reliability report

How did the major AI providers behave this month?

Trailing 30 days: Claude 97.62% uptime across 41 incidents (19 major or critical); OpenAI 99.35% uptime across 27 incidents (7 major or critical).

Numbers come straight from each provider's official Statuspage, recomputed on every visit. The greener cell on a row is the provider that came out ahead; ties get no tint. Hover a row for the metric's definition.

Metric
Claude
OpenAI
  1. Reliability
    1. Composite reliability score
      42.0
      ▲55.4
    2. Uptime
      97.623%
      ▲99.355%
    3. Total downtime
      17h 7m
      ▲4h 38m
    4. Active right now
      0
      0
  2. Incidents
    1. Total incidents
      41
      ▲27
    2. Major + critical
      19
      ▲7
    3. Minor
      17
      17
    4. Longest single incident
      48h 32m
      ▲36h 1m
    5. Longest clean streak
      2d 20hApr 3–6
      ▲
      3d 17hApr 16–20
  3. Time to fix
    1. Median fix time
      ▲1h 8m
      1h 49m
    2. On a bad day
      ▲5h 56m
      12h 20m
    3. Closed vs still open
      39 closedall wrapped up
      26 closedall wrapped up
  4. Patterns
    1. What broke most
      claude.ai32 incidents · 21h 13m attributed
      Conversations11 incidents · 31h 6m attributed
    2. Most common failure type
      Elevated errors23 of 41
      Elevated errors9 of 27
    3. Busiest day of week
      Wednesday10 incidents on this day
      Tuesday7 incidents on this day
    4. Busiest hour of day
      6 AM UTC4 incidents started here
      10 AM UTC4 incidents started here

What this report doesn't cover

  • No synthetic API probing — we only see what each provider publishes on their Statuspage. Quietly degraded responses that never get acknowledged on the status page are invisible here.
  • No regional breakdown — Statuspage feeds don't reliably expose region-level impact, so a US-only outage and a global one count the same.
  • No request-volume weighting — “1 incident” means one acknowledged incident, regardless of how many calls it actually affected.