OpenAI launches Codex-Spark, coding model 15 times faster than GPT-5.3-Codex
Technology
OpenAI just launched Codex-Spark, a new AI model that speeds up coding by a whopping 15 times compared to GPT-5.3-Codex.
Right now, it's available to $200-a-month Pro users in a research preview.
Codex-Spark is all about quick responses and rapid code tweaks, using tech like reduced latency and always-on WebSocket connections to keep things moving fast.
Codex-Spark on Cerebras chip, underperforms GPT-5.3-Codex
Codex-Spark runs on Cerebras's powerful Wafer Scale Engine 3 chip, which helps with the speed boost.
But while it's super quick, it underperforms GPT-5.3-Codex on SWE-Bench Pro and Terminal-Bench 2.0.
Still, this launch is a big step in OpenAI's partnership with Cerebras as they work toward smarter models that balance speed with deeper problem-solving for coders everywhere.