OpenAI launches lightweight coding model on Cerebras chips

On Thursday, OpenAI announced the launch of GPT-5.3-Codex-Spark, a new lightweight AI coding model that functions as a smaller version of its agentic coding tool, Codex. According to OpenAI, the model is designed specifically to achieve faster inference, which is the process of running a trained AI model to make predictions. By optimizing this process, the tool aims to reduce latency—the time delay between a user request and the system response—for developers seeking rapid, real-time collaboration. Ahead of the official release, OpenAI CEO Sam Altman hinted at the product launch, stating that the company had a special release prepared for Codex users on the Pro plan.

To power the new model, OpenAI is utilizing dedicated chips from its hardware partner, Cerebras. Specifically, GPT-5.3-Codex-Spark is powered by Cerebras’ Wafer Scale Engine 3, a large-format processor chip architecture that features 4 trillion transistors. This integration marks a significant step in the physical infrastructure partnership between the two companies, which began with a multi-year agreement announced last month. OpenAI stated that the multi-year contract is worth over $10 billion. At the time of the initial partnership announcement, OpenAI noted that integrating Cerebras into its broader mix of compute solutions was focused on making its artificial intelligence systems respond much faster.

The collaboration represents a milestone in the hardware agreement. Sean Lie, CTO and co-founder of Cerebras, emphasized the potential of the joint effort. “What excites us most about GPT-5.3-Codex-Spark is partnering with OpenAI and the developer community to discover what fast inference makes possible — new interaction patterns, new use cases, and a fundamentally different model experience,” Lie said in a statement.

The launch also highlights the growing market presence of Cerebras. Last week, the hardware company announced that it had raised $1 billion in fresh capital. This funding round established the company’s valuation at $23 billion, underscoring its rising prominence as a provider in the artificial intelligence compute sector.

Why it matters

OpenAI’s release of GPT-5.3-Codex-Spark marks a new level of integration in the company’s physical infrastructure by utilizing Cerebras’ dedicated chips to achieve lower latency and faster inference for coding tasks.