[key points]

Key points

• •OpenAI released a new production AI model, GPT-5.3-Codex-Spark, running on Cerebras hardware instead of Nvidia.
• •This coding model is reported to be 15 times faster than its predecessor, delivering over 1,000 tokens per second.
• •Codex-Spark is available as a research preview for ChatGPT Pro subscribers through various interfaces and offers a 128,000-token context window.