ChatGPT ultra-long chat and load speeds 94%: 741 rounds down from 28 seconds to 2 seconds

OpenAI is preparing to optimize ChatGPT and Codex's super-long conversation experience and go online next week. After an internal test of a 741-wheel, 231MB megachamber, the average opening time dropped from 2762 seconds to 1.66 seconds, increasing the load speed by approximately 94 per cent. This optimization is mainly for loading chat records, not for the speed of model answers. During the test, the number of dialogue entries to be loaded dropped from 15,529 to 64, the number of requests dropped from 894 to 16, and the memory growth of the entire App decreased by 41.2 per cent. For users who use ChatGPT or Codex for long periods of time, this significantly reduces the waiting time for opening old conversations. This time, OpenAI directly reduced the amount of data to be loaded。