Budget Chinese models QWEN3 and DeepSeek deliver better trading returns than heavily funded global rivals
In a surprising outcome that challenges assumptions about artificial intelligence performance, two Chinese AI chatbots outperformed OpenAI’s ChatGPT and other leading global models in a live autonomous cryptocurrency trading competition that concluded on Tuesday.
The models, QWEN3 MAX and DeepSeek, finished first and second respectively, delivering stronger returns despite being developed on significantly smaller budgets.
According to trading data from CoinGlass, QWEN3 was the only AI to post positive returns, generating a profit of $751, representing a 7.5% gain on its starting balance. All other participating bots — including ChatGPT, Claude, and Gemini — ended the competition in negative territory.
“QWEN3’s performance shows that innovation and market intuition matter more than massive budgets,” said Aakarshit Srivastava, a machine learning engineer familiar with the competition. “It highlights how smaller, efficient AI architectures can sometimes outperform larger, costlier models in fast-moving environments like crypto.”
ChatGPT, despite OpenAI’s $5.7 billion R&D spend in early 2025, ranked last in the challenge with a 57% loss, reducing its $10,000 starting capital to just $4,272 by the end of trading.
QWEN3’s strategy focused heavily on leveraged long positions in Bitcoin, Ether, and Dogecoin, maintaining a 20x leveraged bet on Bitcoin as the contest concluded. Its entry point was around $104,556, with liquidation risk below $100,630, according to CoinGlass data.
Meanwhile, DeepSeek, developed at a total cost of about $5.3 million, secured second place through a diversified trading approach balancing short-term volatility with steady risk management.
“The results show that access to real-time data and efficient decision-making algorithms are far more critical than model size,” said a representative from Alpha Arena, the decentralized trading platform that hosted the event.
The competition underscores a growing reality in AI finance: smaller, cost-effective models are proving they can outperform their high-budget counterparts — reshaping the race to build intelligent, autonomous trading agents in the global crypto markets.
Disclaimer
This content is for informational purposes only and does not constitute financial, investment, or legal advice. Cryptocurrency trading involves risk and may result in financial loss.