OpenAI's Jalapeño chip claims 104.3x better efficiency than Nvidia GB300
rohanpaul_ai · x · 2026-08-25
OpenAI claims its new 'Jalapeño' chip delivers 104.3x more throughput per kilowatt than Nvidia's GB300 at matched DeepSeek R1 decoding speeds (12,258 vs. 118 mixed tokens/s/kW).
Key Specs & Details:
- Power Efficiency: Rated at 700W, compared to 1,200W for GB200 and 1,400W for GB300.
- Throughput Gains: Achieves 2.7x, 4.1x, and 3.8x higher per-user decoding throughput on GPT-OSS, DeepSeek R1, and Kimi K2.5 respectively.
- Technical Innovation: Attributes efficiency to keeping KV-cache and model state local, minimizing data movement during inference.
- AI-Accelerated Design: AI helped move from design to tape-out in nine months; some AI-generated kernels ran 1.5–1.8x faster than human expert implementations.
- Timeline: Planned deployment by year-end, with Gen 2 and Gen 3 already in progress.
More from Infra
- engy.ai processes 50B+ tokens in a day as open-model inference demand surges — const_reborn · 2026-10-16
- New inference engine 'e' hits 10,500 tok/s on a single DGX Station — brandon_galang · 2026-10-16
- SGLang ships Day-0 support for FLUX 3 Image Klein: 10 reference images, BF16/FP8/NVFP4 variants — ying11231 · 2026-10-16
- NVIDIA's Nemotron 3.5 Super VL hits CoreWeave: 12B-active multimodal MoE at $0.25/M input tokens — wandb · 2026-10-16
- 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