All model families
gpt-5
Median total latency per provider over the last 24 hours, from live production traffic. This is the signal Requesty's latency router samples when it picks a provider for gpt-5.
Provider latency, last 24 hours
30 minute buckets · UTCCurrent routing standings
gpt-5
1openai↑100.0%10.2s53.5s21%$3.42Win share is simulated with the production router's Thompson Sampling over the last hour of gpt-5 traffic.