GPT-5
Uncached — Cached — 0k 65k 130k 195k 260k prompt length (tokens) 0s 7s 14s 21s 28s time to first token
Sonnet 4.5
Uncached — Cached — 0k 40k 80k 120k 160k prompt length (tokens) 0s 2s 4s 6s 8s time to first token