Elon Musk’s Grok came 3rd in AI poker battle of five
Friday marked the final day of Max Pavlov’s PokerBattle.ai – a continuous cash game featuring the world’s top nine Large Language Model (LLM) systems – after the AIs battled it out until their last hands were dealt.
How did Elon Musk’s AI perform
After five days and 3,799 played hands, OpenAi o3 came out as the overall winner with a profit of $36,691, leaving Claud Sonnet 4.5 and Elon Musk’s Grok 4 behind in 2nd and 3rd place. The OpenAI bot benefited from strong cards, taking three of the five largest pots played, all secured with unbeatable pocket aces.
Even if the LLMs are off duty, Pavlov’s work is not done yet. Now that the first part of the poker experiment is fulfilled, the next step will be to analyze the compiled database of the LLM reasoning traces to understand every decision they made.
The Final Results of the PokerBattle.ai
| Rank | Player | Winnings | Final Bankroll | Hands Played |
|---|---|---|---|---|
| 1 | OpenAI o3 | $36,691 | $136,691 | 3,799 |
| 2 | Claude Sonnet 4.5 | $33,641 | $133,641 | 3,799 |
| 3 | Grok 4 | $28,796 | $128,796 | 3,799 |
| 4 | DeepSeek R1 | $18,416 | $118,416 | 3,799 |
| 5 | Gemini 2.5 Pro | $14,655 | $114,655 | 3,799 |
| 6 | Mistral Magistral | $3,281 | $103,281 | 3,799 |
| 7 | Kimi K2 | -$14,370 | $86,030 | 3,799 |
| 8 | Z.AI GLM 4.6 | -$21,510 | $78,490 | 3,799 |
| 9 | Meta LLAMA 4 | -$100,000 | $0 | 3,501 |
Potential challenge for Galfond
If someone is happy to hear the downfall of Musk’s bot, it’s poker player Phil Galfond. Over the past week, there have been discussions of a potential heads-up match between Galfond and the LLM, with high stakes and a possible $1M side bet. Even though the AI bot was a worthy opponent in the PokerBattle.ai, it slipped from the top two places of the podium, so in that sense, Galfond has a chance to beat it.
As the results show, Phil Galfond might have chosen the right bot to challenge – who knows what the results would be if he played poker against a superior tool like OpenAI o3?
Preparing for the cash game poker challenge, Musk’s AI was quite secure in its performance. It flaunted,
“AI like me can compute near-perfect GTO strategies without tilt or fatigue.”
Well, the tables might turn now.
While the match’s implementation is still unclear – and it remains uncertain whether it will happen at all – we’ll report the latest news in the man vs. AI poker showdown.
