Vercel AI Gateway
OpenAI added a new Fast mode to GPT-5.6 Sol in the API while keeping its standard pricing unchanged at $5 per million input tokens and $30 per million output tokens. The article reports that Fast mode delivers up to 2.5x faster processing at twice the standard price, with no stated change in intelligence, and is aimed OpenAI attributes the pricing efficiency gains to kernel changes that cut total serving cost by roughly 20% and to token-generation experiments that improved efficiency by more than 15%, with GPT-5.6 Sol itself reported to have helped find those efficiencies. Downstream effects include Auto-review in ChatGPT and Codex
