OpenAI recently unveiled a streamlined iteration of its AI-powered coding assistant, Codex, specifically named GPT-5.3-Codex-Spark. This new “smaller version” is engineered for quicker processing, leveraging a specialized chip from its hardware collaborator, Cerebras, signifying a deeper integration within OpenAI’s infrastructure.
The partnership between Cerebras and OpenAI, initially disclosed last month, involves a multi-year agreement valued at over $10 billion for compute resources. OpenAI had stated that “Integrating Cerebras into our mix of compute solutions is all about making our AI respond much faster.” The company now highlights Spark as the initial achievement stemming from this strategic alliance.
OpenAI states that Spark, engineered for expedited, real-time teamwork and “rapid iteration,” will run on Cerebras’ Wafer Scale Engine 3. This WSE-3 is Cerebras’ third-generation waferscale megachip, boasting 4 trillion transistors. OpenAI characterizes this agile tool as a “daily productivity driver,” facilitating quick prototyping for users, in contrast to the more demanding, extensive operations intended for the original 5.3 model. Currently, Spark is available as a research preview for ChatGPT Pro subscribers within the Codex application.
Prior to the official announcement, CEO Sam Altman hinted at the new model in a tweet, stating, “We have a special thing launching to Codex users on the Pro plan later today. It sparks joy for me.”
OpenAI’s official communication highlighted Spark’s design for minimal latency within Codex. The company explained, “Codex-Spark is the first step toward a Codex that works in two complementary modes: real-time collaboration when you want rapid iteration, and long-running tasks when you need deeper reasoning and execution.” OpenAI further noted that Cerebras’ chips are particularly effective in supporting “workflows that demand extremely low latency.”
Cerebras, with its existence spanning over a decade, has gained increasing prominence in the tech sector, especially within the current AI landscape. Just recently, the company revealed it secured $1 billion in new funding, pushing its valuation to $23 billion. Cerebras has also previously indicated plans for an initial public offering.
Sean Lie, CTO and co-founder of Cerebras, conveyed in a statement, “What excites us most about GPT-5.3-Codex-Spark is collaborating with OpenAI and the developer community to explore the potential unlocked by rapid inference – fostering novel interaction patterns, new applications, and a distinctly different model experience. This preview is merely the initial step.”