3 ms·
They did compare it to other models: https://x.com/OpenAI/status/1999182104362668275 https://x.com/OpenAI/status/1999182104362668275 https://i.imgur.com/e0iB8K
by Tiberium 10mo ago
They did compare it to other models: https://x.com/OpenAI/status/1999182104362668275 https://x.com/OpenAI/status/1999182104362668275
https://i.imgur.com/e0iB8KC.png https://i.imgur.com/e0iB8KC.png
- enlyth 10mo agoThis looks cherry-picked, for example Claude Opus had a higher score on SWE-Bench Verified so they conveniently left it out, also GDPval is literally a benchmark made by OpenAI
- minadotcom 10mo agoagreed.
- tobias2014 10mo agoAnd who believes that the difference between 91.9% and 92.4% is significant in these benchmarks? Clearly these have margins of error that are swept under the rug.
- whimsicalism 10mo agouh oh, where did SWE bench go :D
- whimsicalism 10mo agomaybe they will release with gpt-5.2-codex
- sergdigon 10mo agoThe fact that the post is comparing their reasoning model against gemini 3 pro (the "non reasoning" model) and not gemini 3 pro deep think (the reasoning one) is quite nasty. If you compare GPT5.2 thinking to gemini 3 pro deep think, the scores are quite similar (sometimes one is better sometimes the other one is)