On August 25, OpenAI announced the first batch of performance test results for its self-developed AI inference chip, Jalapeño, at the Hot Chips conference. Designed primarily for large language model inference, the chip demonstrated 1.5 to 1.9 times higher performance per watt compared to NVIDIA's GB200 and GB300 systems in tests across multiple publicly available large models, while reducing end-to-end latency by 1.7 to 3.6 times. The Jalapeño chip has a rated power consumption of 700W, with actual sustained power consumption not exceeding 550W in measurements. Currently, the tests are being conducted using the A0 version of the engineering chip, with the next-generation B0 version already in the manufacturing stage. The first-generation chips are set to be deployed in OpenAI's own computing infrastructure by the end of 2026.