GPT-5 just caught them *all* (Grok 4.20 and Gemini 3.0)

The GPT-5 Pokemon stream was awesome. There were cases where GPT (the player) would "one shot" entire puzzles and trainers with multiple Pokemon. It also passed the league using only an under-leveled Charizard (Saladninja) by strategic move choice and item usage.The difference between that and the earlier AIs was night and day. Of course there were a lot of bonks also (if you know you know).

Well, the IOI and IMO medals certainly show that the models haven’t hit a wall.

Im always most hyped about new gemini models. Theyve been consistently good all round, and good cost. Claude has just been overpriced high end above average but not really all purpose

Good stuff Wes! Idea… ask any gpt about ‘recent research on using AI to look at Archaea for possible new antibiotics’….answer is awesome! Maybe work the info into some new episode….

Thanks for the good vids Mr Wes.

Dear Wes, can you please add to description links to the articles you show?

I love all those companies comparison. Strategies were the only post the ones that are behind them.

4:37 the interview was crazy, where he pretended to have come up with the concept of the “singularity”.

Thank you.

Wild how GPT-5 is pulling ahead again have you tried testing it with Genum AI to see how consistent it stays compared to Grok and Gemini across longer runs