OpenAI says its new AI chip, Jalapeño, completes tasks more efficiently and returns responses faster than rival AI systems, according to a blog post published this week. During a briefing with reporters, OpenAI hardware vice president Richard Ho said the chip offers the “best of both worlds” with lower latency and higher throughput, noting that AI systems typically have to make a trade-off between the two.
First introduced in June, Jalapeño is an Application-Specific Integrated Circuit (ASIC) developed in partnership with Broadcom. It is designed for AI inference, the process of running a trained AI model to complete a task or deploy an agent.
How Jalapeño performed in testing
To measure the chip’s performance, OpenAI used InferenceX, a benchmarking platform that shows how well AI systems handle inference. The test compared Jalapeño against the best results recorded at the time, which were achieved with Nvidia’s GB200 or GB300 superchips.
OpenAI says Jalapeño delivered 1.5 to 1.9 times more AI work per watt across GPT-OSS 120B, DeepSeek R1, and Kimi K2.5 1T than the comparison systems. It also offered 1.7 to 3.6 times lower end-to-end latency across the same three models. According to Ho, that means the chip can provide users with faster responses, more responsive agents, and more reliable access as demand grows.
Deployment plans for 2026 and beyond
OpenAI plans to deploy Jalapeño in small volumes by the end of 2026, before beginning to ramp up production into 2027, Ho added. The company has not said how many chips it intends to deploy next year.
Even with these performance improvements, Ho said OpenAI does not expect to replace its entire chip lineup with Jalapeño, describing an overall compute strategy that still includes what he called “very good partners”, such as Nvidia. The company will continue developing the second and third generations of the new chip.
Source
Image: theverge.com
