loading...

THE NINES

Who's chasing 99.999%?

methodology

Claude
OpenAI

Daily Head-to-Head

close

Methodology

Data Sources

Claude status data is pulled from status.claude.com which embeds per-component daily uptime data in a JavaScript variable (uptimeData). OpenAI data comes from their Statuspage API at status.openai.com/api/v2/ plus their Atom feed for incident history.

Components Tracked

  • Claude: Claude API, claude.ai, Claude Code
  • OpenAI: OpenAI APIs, ChatGPT, Codex

Status Colors

  • Green: Fully operational, no incidents
  • Yellow: Degraded performance, incidents reported but no measured downtime
  • Orange: Partial outage under 12 hours
  • Red: Major outage, 12+ hours of partial outage
  • Blue: Scheduled maintenance

Weighted Daily Scoring

Each day is a head-to-head faceoff. Services are weighted by criticality:

  • API + Chat = 3x weight (core products)
  • Code = 2x weight (developer tools)

Per-service daily scores: Operational = 100%, Degraded = 60%, Partial Outage = 30%, Major Outage = 0%, Maintenance = 80%. The weighted sum determines who wins each day. Ties go to neither side.

Limitations

  • Claude incident titles are not exposed via API; only outage duration is shown
  • OpenAI incident data before Feb 3 2026 is limited (Atom feed range)
  • 90-day uptime percentages are from status page scraped data, not real-time calculated
  • Status granularity is per-day, not per-hour

About

The Nines is a fun, independent comparison tool. It is not affiliated with Anthropic or OpenAI. Data is fetched live from public status pages on each page load.