{"id":649759,"date":"2026-08-28T14:47:43","date_gmt":"2026-08-28T14:47:43","guid":{"rendered":"https:\/\/Blockchain.News\/news\/openai-jalapeno-chip-performance-ai-inference"},"modified":"2026-08-28T14:47:43","modified_gmt":"2026-08-28T14:47:43","slug":"openais-jalapeno-chip-outpaces-rivals-in-ai-inference-performance","status":"publish","type":"post","link":"https:\/\/e-bitco.in\/index.php\/2026\/08\/28\/openais-jalapeno-chip-outpaces-rivals-in-ai-inference-performance\/","title":{"rendered":"OpenAI’s Jalape\u00f1o Chip Outpaces Rivals in AI Inference Performance"},"content":{"rendered":"
\n

Iris Coleman<\/a> Aug 28, 2026 14:47<\/span> <\/p>\n

OpenAI’s Jalape\u00f1o custom AI chip achieves up to 1.9x better throughput per kilowatt and 3.6x lower latency than competing systems, redefining efficiency in AI inference.<\/p>\n

\"OpenAI's <\/a> <\/figure>\n

OpenAI has unveiled benchmark results for its first custom AI inference chip, Jalape\u00f1o, showcasing industry-leading performance in efficiency and speed. According to OpenAI\u2019s data, Jalape\u00f1o delivered between 1.5 to 1.9 times higher throughput per kilowatt and up to 3.6 times lower latency when compared to NVIDIA\u2019s GB300-class systems. These gains highlight a significant leap in AI inference capabilities, particularly for large language models (LLMs) like GPT-OSS 120B, DeepSeek R1, and Kimi K2.5 1T.<\/p>\n

Jalape\u00f1o was first announced on June 24, 2026, as a collaboration with Broadcom, marking OpenAI\u2019s move into custom silicon development. Unlike general-purpose GPUs, Jalape\u00f1o is purpose-built for running already-trained AI models, such as those powering ChatGPT. The chip is optimized specifically for inference workloads, minimizing power consumption and reducing response times. While NVIDIA\u2019s GPUs dominate the broader AI hardware market due to their programmability and ecosystem, Jalape\u00f1o\u2019s targeted design offers OpenAI a competitive advantage for its internal needs.<\/p>\n

The benchmarks, released on August 25, highlight Jalape\u00f1o\u2019s ability to handle interactive AI workloads with unprecedented efficiency. For example, on the largest tested model, Kimi K2.5, the chip achieved 1.5 times higher peak performance per watt and reduced end-to-end latency by 3.4 times compared to the competition. These metrics underscore its capability to process high-demand tasks, such as real-time chatbot interactions, with reduced energy costs \u2014 a crucial factor as AI adoption scales globally.<\/p>\n

OpenAI\u2019s engineering team credited AI itself for accelerating Jalape\u00f1o\u2019s development. Leveraging internal AI tools, the team moved from design to tapeout in just nine months. Additionally, AI played a direct role in optimizing the chip\u2019s circuits and programming, resulting in faster deployment and improved performance. Notably, AI-generated implementations for specific model blocks outperformed human-written code by 1.5 to 1.8 times, further streamlining development cycles.<\/p>\n

Though Jalape\u00f1o is not available for external sale, its impact on OpenAI\u2019s operations could be profound. Faster, more power-efficient inference allows the company to lower costs and serve more users, improving its operating leverage. With Gen 2 and Gen 3 chips already in development, OpenAI is doubling down on custom silicon as a strategic advantage.<\/p>\n

In the broader market, Jalape\u00f1o\u2019s performance positions OpenAI as a potential competitor to Google\u2019s Tensor Processing Units (TPUs), though the two differ in scope. Google TPUs cater to both training and inference and are available commercially via Google Cloud, while Jalape\u00f1o is strictly an internal tool for inference. The comparison to NVIDIA is equally nuanced: while NVIDIA GPUs remain the go-to for versatility and ecosystem support, Jalape\u00f1o\u2019s specialization offers superior efficiency for specific workloads.<\/p>\n

Looking ahead, OpenAI plans to deploy Jalape\u00f1o at scale within its compute infrastructure by the end of the year. As the company continues to refine the platform and expand its capabilities, Jalape\u00f1o represents a key step in meeting the growing global demand for AI-powered applications while managing costs and environmental impact.<\/p>\n

Image source: Shutterstock<\/i><\/span> Source<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Iris Coleman Aug 28, 2026 14:47 OpenAI’s Jalape\u00f1o custom AI chip achieves up to 1.9x better throughput per kilowatt and 3.6x lower latency than competing systems, redefining efficiency in AI inference. OpenAI has unveiled benchmark results for its first custom AI inference chip, Jalape\u00f1o, showcasing industry-leading performance in efficiency and speed. According to OpenAI\u2019s data, […]<\/p>\n","protected":false},"author":1,"featured_media":649760,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12],"tags":[21143,21786,26403,25,8513],"class_list":{"0":"post-649759","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-blockchain","8":"tag-ai-hardware","9":"tag-ai-inference","10":"tag-jalapeno-chip","11":"tag-news","12":"tag-openai"},"_links":{"self":[{"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/posts\/649759","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/comments?post=649759"}],"version-history":[{"count":0,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/posts\/649759\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/media\/649760"}],"wp:attachment":[{"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/media?parent=649759"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/categories?post=649759"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/e-bitco.in\/index.php\/wp-json\/wp\/v2\/tags?post=649759"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}