OpenAI's In-House Jalapeño Chip Reportedly Beats Nvidia on Efficiency

OpenAI has taped out its in-house AI-designed Jalapeño chip in about 9 months, claiming higher tokens-per-watt than Nvidia's GB200 on models like GPT-OSS 120B and DeepSeek R1, with deployment planned before year-end.

2026-08-30 ~ 2026-08-30 · 3 related posts

Full story(6 episodes)→

1 near-duplicate retellings: Beth_Kindig