A new version of OpenAI’s Codex is powered by a new dedicated chip

OpenAI’s latest Codex model, dubbed Codex-Spark, runs on dedicated Cerebras chips, marking the company’s first major hardware partnership beyond Nvidia. The chip-accelerated model generates code roughly fifteen times faster—over 1,000 tokens per second—and supports a 128,000-token context window with text-only inputs, now accessible to ChatGPT Pro subscribers.

本文综合以下全部报道写成，而非依据任何单一来源。