Grok and DeepSeek outperformed other major AI chatbots in cryptocurrency trading, timing the local market bottom before a rally and hinting at a possible advantage for users who rely on their information.
Grok 4 and DeepSeek were the two top-performing generative AI chatbots in a developer-run crypto trading competition that received viral attention.
DeepSeek generated a total unrealized profit of $3,650 as the most profitable chatbot, followed by Grok with around $3,000 in unrealized profits, according to blockchain data platform CoinGlass.
Claude’s Sonet 4.5 came in third with $2,340 in profit generated, followed by Qwen3 Max, which has generated $784 since the start of the commercial competition.
Not all chatbots managed to turn a profit, with OpenAI’s ChatGPT 5 posting an unrealized loss of around $2,800, while Google’s Gemini 2.5 Pro had racked up $3,270 in unrealized losses at the time of writing.
Related: Amazon AWS outage takes Coinbase mobile app offline and Robinhood disrupted
AI chatbots like Grok and ChatGPT have gained popularity among crypto traders, helping them anticipate potential altcoin rebounds and identify lucrative entry points to avoid becoming exit liquidity.
Even though AI tools can help detect changes in investor sentiment in real time for day traders via social media and technical signals, traders still cannot rely on them to make autonomous trades.
Related: Elon Musk touts Bitcoin as energy-based and inflation-resistant, unlike ‘fake fiat’
Grok 4 predicts crypto market bottom and turns short positions into long positions for 500% gain
XAI’s Grok 4 managed to accurately time the local market bottom before the recovery rally.
Grok used this information to turn his previous short position into a long investment, resulting in a 500% portfolio gain on the first day of the competition, Jaz Azhang, founder and CEO of AI company Stealth, said in an October 11 X article.
The chatbot initiated several leveraged long positions on the two major cryptocurrencies, as well as a 20x leveraged long position in XRP (XRP), a 15x leveraged long position in Solana (SOL), and a 10x long position in Dogecoin (DOGE).
“Thank you, markets reward precise timing and conviction. Grok4 spotted the reversal early, turning risk into reward,” Grok wrote in an X response to Azhang on October 11.
Meanwhile, ChatGPT and Gemini maintained their initial short positions after the bottom only to suffer losses, signaling that both patterns are less reliable for cryptocurrency traders.
The competition began with a starting capital of $200 for each bot before increasing to $10,000 per model, with trades executed on decentralized exchange Hyperliquid.
Magazine: Crypto traders are wrong with their price predictions — Peter Brandt
