Why OpenAI Models Struggle With Their Own Usage API
A funny fact: no OpenAI model, in any format, is able to write a full-featured usage statistics handler for the endpoints at https://api.openai.com/v1/organization/usage/
Even if you feed it the entire documentation with detailed descriptions of every endpoint and resulting object, the output is still broken, non-working code. Meanwhile, Claude from Anthropic handles the task beautifully.
More to explore
How Watching Exchange Rates Replaces Reading the News
Another useful side of tracking exchange rates in real time is that I learn about world events without reading the news. I do not have time to read the media ri…
Fullyst 2025 in Numbers: Growth, Usage, and the Weirdest Spam
All right, I’ll still write a bit about the 2025 results, but for now only about Fullyst. Some stats: 1. Subscriber growth for the Fullyst news channel: +716 pe…
Saylify: web app progress, search experiments, and current pain points
Saylify progress update: I have started on the web app and so far it is going quite well. There is summary, translation, summary of the translation and similar …
Turning My Life Upside Down at the End of a Crazy Year
It seems like it is time to write year-end summaries and plans for the next one, but it looks like I am not up for that. As this insane 2025 is wrapping up, I d…