NO, AI PROGRESS IS NOT GRINDING TO A HALT

Summary from the TLDR Founders newsletter

Everyone's declaring GPT-5 a disappointment that proves AI progress is hitting a wall, but they're measuring it wrong. When GPT-4 launched in 2023, it was compared to technology from 2020 - of course, it looked revolutionary. Now GPT-5 is being compared to models released last month. If GPT-4 came out today, it would be dead on arrival. It handled 8,200 tokens versus GPT-5's 400,000, cost 10x more to run, and scored 25/100 on benchmarks where GPT-5 gets 69. On real-world coding tests, GPT-4 solved 2.8% of problems while GPT-5 solved 65%. GPT-4 can handle 5-minute human tasks - GPT-5 can handle tasks that are more than two hours. However, because everyone expected another ChatGPT moment and OpenAI bungled the release, the narrative has solidified in DC that AI is stalling.