OpenAI has cut inference costs for its AI models by more than half, according to a report.

The optimizations were applied to ChatGPT, reducing the number of Nvidia GPUs needed to just a few hundred at times.

AI-generated: summaries written by AI from the linked sources. How we use AI