Accelerating GPT-5.6 Sol Ultrafast
Summary
The article announces Ultrafast Mode for GPT-5.6 Sol and claims up to 750 output tokens per second with no quality tradeoff for time-sensitive workloads. It benchmarks Ultrafast against other models and highlights the wafer-scale engine with 44 GB of on-chip SRAM, noting limited preview availability and potential security and uptime benefits.