rki.news | Source: Anadolu
SAN FRANCISCO, August 25: OpenAI said Tuesday that its first custom artificial intelligence chip, Jalapeno, outperformed Nvidia systems in company tests, delivering higher energy efficiency and faster responses.
The ChatGPT developer said Jalapeno produced 1.5 to 1.9 times more AI work per watt at peak throughput than comparison systems, while reducing end-to-end latency by 1.7 to 3.6 times.
For highly interactive workloads, the chip delivered 2.1 to 4.1 times higher performance, according to OpenAI.
Tests used the InferenceX benchmark and OpenAI’s GPT-OSS 120B, DeepSeek R1 and Moonshot AI’s Kimi K2.5 models. Comparisons included Nvidia’s GB200 and GB300 systems, but not its newer Vera Rubin generation.
Jalapeno is designed specifically for AI inference rather than training. It is rated at 700 watts, while sustained consumption remained at or below 550 watts during testing.
Developed with Broadcom, the chip reached tapeout in nine months. OpenAI plans to deploy it by the end of 2026, while continuing to use Nvidia and other partners’ hardware.
The company is already developing second- and third-generation processors.