OpenAI's Jalapeño chip tops benchmarks with higher throughput and efficiency

TechCrunch AI · rss · 2026-08-25

According to Semianalysis's InferenceX benchmark, OpenAI's custom chip Jalapeño demonstrates superior inference performance. It achieves both higher tokens per user and greater throughput per kilowatt compared to the current state-of-the-art available alternatives.

Original post →

More from Infra

Infra channel →