Not badX Freeze (@XFreeze):I tested the same prompt on both Grok 4.3 and GPT 5.5:“Count to 10 starting from 11”ChatGPT 5.5 gave the obvious 11–20Grok 4.3 gave 11, 10 and explained why going backwards was the only logical moveGrok’s logical reasoning is at a level most models still can’t even touch https://t.co/HtvAVHOVxd ...

Finally, here are 10 more evaluations I ran using DeepEval on logical reasoning tasks.- GPT-5 won in 2 cases.- Grok 4 won in 3 cases.- A Tie happended in 5 cases.Grok 4 was found to be better in terms of depth of analysis.Check this👇 https://t.co/4siD5PqJPQ ...