#openai these GPT-5 coding demos are lame. Next time show side by side so we can compare speed and aesthetics by ourself compared to the previous best model. They are not showing anything in these coding demos that we haven't already seen for a year now. Study how cooking informercials work (async work, do some work ahead of time). Get the demo running on two computers at the same time. Then while you're explaining the new changes in this model it can be doing the inference in the background. Then you can come back to check on the results as soon as the model is done.