Gpt 5 output capacity issue

[grim stag]

Hi everyone,

I’m running into an issue with GPT5 while automating repetitive processes in my game code. It seems like I’m asking it to output too many tokens at once, which is causing a connection error. I have already tried splitting the output into multiple parts and having it generate a downloadable file, but when I do this, it often goes off topic from my original prompt. Does anyone know a reliable way to bypass this token limit issue? Or are there any alternative AI tools you’d recommend for handling large code outputs? – Already tried Claude, do not want to pay for a $200 sub.

[tawny shore]

I have the same issue. From one comment to the next even. It just forgets what it’s even said in a thread. I will ask it for help with coding on something in a longer thread and it will say okay, here’s what we’ll do and in the next have no idea what it’s said. You get 125,000 tokens per convo with paid which is an insane amount. It shouldn’t run out unless you’ve written like 5 novels. I have no idea why it’s so confused. I’ve tried taking smaller asks into a different convo and get better requests but for larger complicated ones that need backstory, it’s not helpful anymore. If you get a solution on another platform, I’d be interested.