Скачать или смотреть Checkmate: 2600 ELO Theory, 900 ELO Blunder: Gemini vs. GPT-5.2

Google Gemini (White) vs. OpenAI GPT-5.2 (Black). A match that exposes the biggest weakness of Large Language Models in Chess.

The Highlight: We compare ChatGPT's internal confidence (it thought it was winning!) against Gemini's harsh reality check.

While GPT was writing paragraphs about its "Grandmaster strategy," it missed a simple one-move tactic (14…Qxd4+??) that lost the game instantly. Gemini's analysis punishes this severely, rating the Black pieces at a mere 900-1100 ELO for this game due to the tactical blindness, despite the perfect opening.

In this video:

The "Dragon-ish" Sicilian Opening.

The fatal "Hallucination" Move.

Gemini's Verdict: Why intention doesn't matter if you hang your Queen.

Subscribe for more AI Chess Battles!