OpenAI's GPT-5.6 Sol Achieves 14x Speed Increase with Ultrafast Mode

OpenAI’s GPT-5.6 Sol now operates at a maximum output speed of 750 tokens per second with the newly launched Ultrafast mode. This mode offers a 14x speed increase compared to previous configurations, enhancing the model’s performance significantly. The Ultrafast mode is powered by a $10 billion partnership with Cerebras, which provides the necessary hardware. Notably, this new mode introduces a three-tier pricing structure that monetizes inference speed as a distinct product, allowing developers to optimize costs based on their speed requirements. This development follows OpenAI’s recent advancements in AI agent solutions, indicating a trend towards faster, more efficient AI capabilities. For practitioners, the enhanced speed can lead to more responsive applications and improved user experiences, particularly in real-time scenarios. More details can be found in The Decoder.