OpenAI Jalapeño Outperforms Nvidia Rubin with Massive Software Gains in 5 Days
teortaxesTex · x · 2026-08-26
Analysis by SemiAnalysis reveals OpenAI's custom Jalapeño chip surpasses Nvidia's Vera Rubin in throughput and performance-per-watt, even without Prefill/Decode disaggregation or speculative decoding used by Rubin. This suggests a more balanced design less reliant on P/D disagg. Additionally, the 5-day gap between performance curves highlights OpenAI's rapid software iteration speed.
Related event: OpenAI Claims Custom Inference Chip Jalapeño Beats Nvidia Blackwell(34 posts)→
More from Infra
- Open Source vs Labs: Startups Must Build the Full Stack — matt_slotnick · 2026-08-26
- Next AI hardware race might be about inference speed — Delicious-Flan88 · 2026-08-26
- NVIDIA Releases Get Started Guide for Open Model Routing — NVIDIAAI · 2026-08-26
- AC2 launches private beta: Build a model factory to train, serve, and improve models — ypatil125 · 2026-08-26
- Youdotcom Web Search Tops Benchmark with 74 Score in 18 Seconds — RichardSocher · 2026-08-26
- M6 Mac Mini starts at $899 — andrejusb · 2026-08-26