OpenAI has launched a lighter version of its programming tool, Codex.
GPT-5.3-Codex-Spark is now in research preview.
— OpenAI (@OpenAI) February 12, 2026
You can just build things—faster. pic.twitter.com/85LzDOgcQj
GPT-5.3-Codex-Spark is positioned as a scaled-down version of the GPT-5.3 Codex model introduced in February. It is designed for faster inference, utilizing a specialized chip from its hardware partner, Cerebras.
In January, the companies announced their collaboration.
“Integrating Cerebras into our computing solutions aims to significantly speed up AI response times,” OpenAI stated at the time.
The company refers to the Spark chip as the "first milestone" of their partnership.
This processor is designed for rapid real-time performance, based on the Wafer Scale Engine 3, Cerebras' third-generation mega chip, which features four trillion transistors.
OpenAI describes the new tool as a "daily productivity driver" that aids in rapid prototyping. The original GPT-5.3 Codex is intended for longer, more complex tasks.
In an official statement, OpenAI emphasized that Spark is designed for minimal latency in Codex.
“Codex-Spark is the first step towards Codex, which operates in two complementary modes: real-time collaboration for quick iterations and extended tasks requiring deep reasoning,” the company noted.
Recall that in February 2026, OpenAI released a standalone application for the Codex programming assistant.