On Thursday, OpenAI announced the release of a lightweight version of its agent coding tool Codex. This is the latest model that OpenAI launched earlier this month. The company says GPT-5.3-Codex-Spark is a “smaller version” of that model, designed for faster inference. To power its inference, OpenAI introduced specialized chips from hardware partner Cerebras, enabling a new level of integration in the company’s physical infrastructure.
The partnership between Cerebras and OpenAI was announced last month, with OpenAI announcing it had reached a multi-year agreement with the company worth more than $10 billion. “Integrating Cerebras into our suite of computing solutions means significantly increasing the response speed of our AI,” the company said at the time. OpenAI is now calling Spark the “first milestone” in that relationship.
OpenAI says Spark is designed for rapid real-time collaboration and “rapid iteration” and will be powered by Cerebras’ Wafer Scale Engine 3. WSE-3 is Cerebras’ third-generation wafer-scale megachip, containing 4 trillion transistors. OpenAI describes this new lightweight tool as “an everyday productivity powerhouse that helps users quickly prototype,” rather than the long, heavy tasks that the original 5.3 was designed for. Spark is currently enjoying a research preview for ChatGPT Pro users in the Codex app.
CEO Sam Altman seemed to hint at a new model in a tweet ahead of the announcement. “Later today, we will be launching a special offering for Codex users on the Pro plan,” Altman tweeted. “It causes joy for me.”
In an official statement, OpenAI emphasized that Spark is designed to keep latency as low as possible on Codex. “Codex-Spark is the first step toward Codex operating in two complementary modes: real-time collaboration when you need rapid iteration, and long-running tasks when you need deeper inference and execution,” OpenAI shared. The company added that Cerebras’ chips are excellent at supporting “workflows that require extremely low latency.”
Cerebras has been around for more than a decade, but it is playing an increasingly important role in the technology industry in the AI era. Just last week, the company announced it had raised $1 billion in new funding at a valuation of $23 billion. The company previously announced its intention to pursue an IPO.
“What we’re most excited about about GPT-5.3-Codex-Spark is partnering with OpenAI and the developer community to discover what fast inference can enable: new interaction patterns, new use cases, and fundamentally different model experiences,” Sean Lie, CTO and co-founder of Cerebras, said in a statement. “This preview is just the beginning.”
tech crunch event
boston, massachusetts
|
June 23, 2026
