OpenAI Launches GPT-5.3-Codex-Spark For Real-Time Coding

OpenAI announced on January 14 a partnership with Cerebras and, four weeks later, launched GPT-5.3-Codex-Spark, an ultra-fast model for real-time coding. The smaller, text-only variant includes a 128k context window and OpenAI claims roughly 1,000 tokens per second, prioritizing latency over top-end quality. The release is aimed at interactive developer workflows, enabling faster iteration and rapid prototyping.