OpenAI's Ultrafast Mode Goes Live

🔥 What happened OpenAI is gearing up for a broad rollout of its "Ultrafast" API mode around DevDay on September 29. A hidden speed selector — Standard, Fast, Ultrafast — has already surfaced in the API docs. 💡 Why it matters OpenAI previewed Ultrafast at up to 750 output tokens per second, roughly 14× faster than Standard, powered by Cerebras. Developers could soon pick latency per workload, reserving pricey inference for jobs where speed actually drives revenue. ⚡ Our take This isn't a feature, it's a pricing decision. Teams that default to Ultrafast will burn budget on workloads where 200 ms never mattered.