Grok and DeepSeek have outperformed other major artificial intelligence (AI) chatbots in cryptocurrency trading, timing the market’s local bottom before a recovery rally and hinting at a possible edge for users who rely on their insights. Grok 4 and DeepSeek were the two best-performing generative AI chatbots in a crypto trading competition launched by developers that received viral attention.
DeepSeek generated a total unrealized profit of $3,650 as the most profitable chatbot, followed by Grok with about $3,000 in unrealized profits, according to blockchain data platform CoinGlass.
Claude’s Sonet 4.5 came third with $2,340 in generated profit, followed by Qwen3 Max, which generated $784 since the start of the trading competition. Not all chatbots managed to generate a profit, with OpenAI’s ChatGPT 5 seeing an unrealized loss of about $2,800, while Google’s Gemini 2.5 Pro had also amassed $3,270 worth of unrealized losses at the time of writing.
AI chatbots like Grok and ChatGPT have gained popularity among crypto traders, helping them anticipate potential altcoin rallies and identify lucrative entry points to avoid becoming exit liquidity. While AI tools can help spot real-time investor sentiment shifts for day traders via social media and technical signals, traders still can’t rely on them for autonomous trading.
Grok 4 Predicts Crypto Market Bottom and Flips Positions for Significant Gain
XAI’s Grok 4 managed to accurately time the market’s local bottom before the recovery rally. Grok used these insights to flip its previous short position into a long investment, resulting in a 500% portfolio gain within the first day of the competition, said Jaz Azhang, founder and CEO of AI company Stealth, in an Oct. 11 X post.
The chatbot initiated multiple leveraged long positions on the two leading cryptocurrencies, along with a 20x leveraged long on XRP, a 15x leveraged long on Solana, and a 10x long position on Dogecoin.
“Thanks—markets reward sharp timing and conviction. Grok4 spotted the reversal early, turning risk into reward,” wrote Grok in an X response to Azhang on Oct. 11. Meanwhile, ChatGPT and Gemini maintained their initial short positions after the bottom to incur losses, signaling that the two models are less reliable for cryptocurrency traders.
The competition began with $200 in starting capital for each bot before increasing to $10,000 per model, with trades executed on decentralized exchange Hyperliquid.