OpenAI volatility increased; Anthropic stable
Weekly reliability report for AI providers.
P95 latency
Error rate
Rate limit
Weekly summary
Error rates rose for OpenAI chat endpoints while Anthropic maintained consistent latency and availability across regions.
Top incidents
- OpenAI Elevated Errors
- Groq Partial Outage
Reliability winners
- Anthropic Claude 3.5 Sonnet
- Azure OpenAI GPT-4o
Reliability losers
- OpenAI GPT-4o
- Groq Hosted Llama
Methodology
Rankings blend availability, error rate, latency stability, and rate-limit pressure. For details, see the full methodology.
Read methodology ->Share this report
Week ending 2025-12-27 00:00:00 UTC - OpenAI volatility increased; Anthropic stable