Running OpenAI o3 high (tuned) for this benchmark costed $1000 *per task*, but the results are unbelievable. Although still not AGI.   Source: https://x.com/mikeknoop/status/1870172132136931512   For those who don’t know, the current free version of ChatGPT is a very dumbed down (and inexpensive) version of GPT-4o.