Woofun AI reports that OpenAI engineers have developed a suite of model optimization techniques. These innovations reduce model inference costs by over 50% while decreasing reliance on NVIDIA GPUs. The company may utilize these savings to lower API service prices or increase user query limits for products such as ChatGPT.