Filters
Filtered by tag: inference × Clear
AI Infrastructure Race Accelerates As OpenAI Unveils Jalapeño Chip Results And Redefines Agentic Product Design
Matched: AI, ChatGPT
OpenAI unveiled benchmark results for its Jalapeño inference chip at Hot Chips, showing it outperforms Nvidia's Blackwell configuration on tokens per user and throughput per kilowatt. Developed with Broadcom, the chip targets lower latency and reduced data movement, with limited deployment expected by end of 2026. Alongside the hardware news, OpenAI detailed ChatGPT Work, a $20/month agentic product extending automation beyond coding into general white-collar workflows.