According to CoinMeta, OpenAI has announced the latest test results of its self-developed AI inference chip Jalape ño (translated as: Mexican chili). The chip outperformed NVIDIA GB200 and GB300 super chips in InferenceX benchmark tests. On models such as GPT-OSS 120B, DeepSeek R1, and Kimi K2 .5 1T, it achieved 1.5 to 1.9 times the performance per watt of its competitors, while reducing end-to-end latency by 1.7 to 3.6 times. OpenAI stated that Jalape ño utilizes a ASIC architecture developed in collaboration with Broadcom, specifically designed for AI inference purposes, aiming to increase throughput and reduce latency simultaneously. OpenAI plans to deploy the chip on a small scale by the end of this year, expand production in 2027, and continue to develop second and third-generation chips. However, the company emphasizes that it will not completely replace chips from partners such as NVIDIA.