OpenAI has revealed new benchmark results for its custom-built AI chip, Jalapeño, at the Hot Chips conference in Palo Alto, California, claiming the processor delivers quicker AI responses while consuming less power.
According to the company, Jalapeño handled between 1.5 and 1.9 times more AI workload per unit of energy, and cut response generation time by 1.7 to 3.6 times, based on tests run across three different AI models.
For highly interactive use cases, OpenAI said the chip’s performance edge grew even further, ranging from 2.1 to 4.1 times higher than existing benchmarks.
