OpenAI's Jalapeño Chip Beats Nvidia Blackwell in Efficiency
rohanpaul_ai · x · 2026-08-30
Semianalysis reports that OpenAI's custom inference chip 'Jalapeño' significantly outperforms competitors. At 100 tok/s, it delivers 11M tok/s/MW, roughly 2x the efficiency of comparable Blackwell configs. Remarkably, this is achieved without Multi-Token Prediction (MTP) or speculative decoding, while competing chips utilize MTP.
More from Infra
- Chip equipment beats bull case: 2026 WFE seen at $153B, $278B by 2030 — BenBajarin · 2026-10-16
- Steam now launches in 5.23s, beating Epic's 5.47s — and vLLM gets roasted for slow startup — AlpinDale · 2026-10-16
- Shadeform to Announce Six New Features on Oct 19: Lower GPU Costs, More Suppliers — _jaydeepkarale · 2026-10-16
- Fastino launches Training API for SFT/RL, claiming 25-85% cost cut vs Tinker — kalyan_kpl · 2026-10-16
- Dallas Fed: AI data center debt issuance equals ~1/8 of Treasury duration supply — _onionesque · 2026-10-16
- Running Wan2.1 14B FP16 on a 12GB laptop GPU: VRAM swap stress test finishes in 4m11s — RYuu-1120 · 2026-10-16