Back to all articles
•Technology

OpenAI and Cerebras Collaborate for Faster AI with New Codex Model

View original source

On Thursday, OpenAI announced the release of GPT-5.3-Codex-Spark, a lightweight version of its agentic coding tool. This new model is a "smaller version" optimized for faster inference, made possible by a dedicated chip from Cerebras. This marks enhanced integration in OpenAI's physical infrastructure.

  • Integration with Cerebras: OpenAI's partnership, announced last month, involves a multi-year, $10 billion agreement with Cerebras for advanced compute capabilities aimed at faster AI responses.
  • Technical Specifications: Codex-Spark will be powered by Cerebras' Wafer Scale Engine 3, featuring 4 trillion transistors for real-time collaboration and rapid iteration.
  • Availability: Currently in a research preview for ChatGPT Pro users, Codex-Spark is aimed at enhancing daily productivity through rapid prototyping.
  • Sam Altman's Hint: OpenAI's CEO hinted at the release of this model in a tweet, highlighting its user appeal.
  • Cerebras' Rising Role: Founded over a decade ago, Cerebras continues to gain prominence, having recently raised $1 billion and being valued at $23 billion, with future plans for an IPO.

Impact and Future: This development paves the way for new interaction patterns and use cases in AI modeling, promising a breakthrough in AI's usability.

Note on Tech Event: Promotional details about the TechCrunch Founder Summit 2026 were also included, focusing on growth and industry networking.