14 Feb 2026
GPT-5.3-Codex-Spark: 1,000 Tokens/s on Cerebras Chips
(01)OpenAI launches GPT-5.3-Codex-Spark on Cerebras Wafer Scale Engine 3 — over 1,000 tokens per second, 80% faster roundtrips, and the first non-Nvidia production model. What it means for developers.
OpenAI