I tried it in an 8h coding session. It performs worse than Claude 4 for me, and it’s slower. It made me wait for 10 minutes at a time. Eventually, I gave up, used my brain to understand the problem properly, and hand held Claude to a solution, which took about 1h. I think I lost most of my time with GPT-5 in loops where it fixes one thing at the expense of another while the general approach looked sufficiently sound to fool me but not sufficiently sound to eventually work in all cases.

This might be due to high traffic and not because the model is worse. I’ll give it another chance when the hype fades.

Cloudflare introduces a private beta to a service where engines are required to pay to access the information on a site. It made me think.

AI breaks the open web model in at least three different ways.

• First is that the open web gets filled with AI-generated garbage
• Second is that any word posted anywhere, from websites to DMs, may be used to train models, and then later retold and sold as AI
• Third is that the lack of transparency of how the models work is a fertile ground spreading precisely controlled lies (in the shape Generative Engine Optimization – GEO)

While the first and third don’t bother me much yet, the second bothers me a lot. I feel like Google broke the pact it made with the Internet to provide neutral web search in exchange for profiting from paid search. Now website owners need to pay so their content is crawled, so the different AI tools can present it as universal knowledge. Useful or not, many AI tools feed from the web and give content creators nothing in return. Here’s Cloudflare’s product idea – to charge crawlers, make it less unfair.

However, even though this is a step in the right direction, it still doesn’t feel right, even if there was a way to enforce it.

When BMW designs a car, they charge per car sold, not per car design stolen.