AITom's Hardwareabout 2 hours ago
OpenAI Jalapeño ASIC Beats Nvidia GB300
4 min read
OpenAI's Jalapeño inference ASIC, co-developed with Broadcom, outperforms Nvidia's GB200 and GB300 rack systems, delivering 1.5x to 1.9x more throughput per kilowatt and 1.7x to 3.6x lower latency on SemiAnalysis's InferenceX suite. The 700W chip beat accelerators rated at 1,200W and 1,400W.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
