OpenAI reveals 'Jalapeño' chip specs at Hot Chips, comparisons questioned
AccBalanced · x · 2026-08-27
OpenAI revealed the first real specs for its in-house chip 'Jalapeño' at Hot Chips. While some claim it beats Nvidia, analysts note OpenAI compared it against older Nvidia systems (HBM3e-based Blackwell) rather than the contemporary Rubin (which also uses HBM4).
Key metrics:
- Power: Rated 700W, staying under 550W in tests; comparators normalized to 1.2kW (GB200) and 1.4kW (MI355X).
- Performance: Delivered 1.5x-1.9x more work per watt and 1.7x-3.6x lower end-to-end latency.
- Memory: 216 GiB HBM4 per package at 15.4 TB/s.
Critics highlight the lack of controlled, same-generation benchmarks.
Related event: OpenAI's Jalapeño Chip Debuts at Hot Chips Amid Benchmark Debate(2 posts)→
More from Infra
- Dual RTX 3090s run ComfyUI video gen 2.7x faster via new open-source Windows NCCL backend — unjusti · 2026-08-27
- Nvidia's Zero China Revenue Guidance Reveals Complete Domestic Tech Stack Shift — heyshrutimishra · 2026-08-27
- Freebots Launches Free Domain and Hosting for AI Agents — Daniel_Farinax · 2026-08-27
- ik_llama.cpp monthly update: DSpark speculative decoding, Vulkan IQ4 support, and more — pmttyji · 2026-08-27
- Test: OX Alpha runs on WebGPU for just $0.016 — yuwen_lu_ · 2026-08-27
- Local AI trade-off: 96GB mixed RAM vs. speed — QuirksNFeatures · 2026-08-27