loading...

THE NINES

The reliability race, tracked daily.

Claude
OpenAI

Daily Head-to-Head

Methodology

Data

90-day snapshot from status.claude.com and status.openai.com. Historical bars mirror each provider’s published daily uptime view; active current-day incidents are escalated live from provider summary APIs.

Components

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

Status Colors

  • Green: Operational
  • Yellow: Degraded, no measured downtime
  • Orange: Partial outage (<12h)
  • Red: Major outage (12h+)
  • Blue: Maintenance

Scoring

Each day is a weighted head-to-head. API and Chat carry 3x weight, Code carries 2x. Per-service scores: Operational 100%, Degraded 60%, Partial 30%, Major 0%, Maintenance 80%. Highest weighted sum wins the day.

Ties

Equal weighted scores = tie, no winner. Most ties are days where every service on both sides is fully operational (both score 8/8).

Limitations

  • Dates follow the providers’ own daily uptime calendars, which are UTC-based
  • Standings and summary metrics compare the tracked race services, not every provider component
  • OpenAI incident titles are attached per tracked service using official affected-component metadata from the public status feed

About

Independent project, not affiliated with Anthropic or OpenAI.