Alibaba’s Qwen3-Max Tops Crypto Trading Challenge Against Five AI Models Including GPT-5, Musk's Grok

The results are from Nof1’s Alpha Arena, a benchmark designed to measure AI's investing abilities.

• With $10,000 and two weeks' time, Qwen3-Max delivered 22.32% returns, the highest in the pack.
• DeepSeek’s V3.1 returned 4.89%, while the remaining four models, all from U.S. companies, recorded losses.
• The results are from Nof1’s Alpha Arena, a benchmark designed to measure AI's investing abilities.

Alibaba’s Qwen3-Max model made the best returns in crypto trading in a contest pitting six U.S. and Chinese models against each other, according to a report in the South China Morning Post.

In the first round of testing on U.S. research firm Nof1’s Alpha Arena, a live trading experiment that concluded on Tuesday, the Qwen3-Max model generated a 22.32% return on an initial investment of $10,000 over two weeks.

The goal of the process was to benchmark how AI models perform in actual market conditions, including decision-making, risk management, timing, and position sizing. The models were given only quantitative market data and no access to news.

DeepSeek’s V3.1 Chat model, also from China, delivered 4.89% gains, while all four U.S. models – from OpenAI, Anthropic, Google DeepMind, and Elon Musk’s xAI – recorded heavy losses, according to the SCMP report (SCMP is owned by Alibaba). OpenAI’s GPT-5 fared worst, dropping 62.66%.

Nof1 in a blog post cautioned that the early results “may be the result of luck”, adding that future rounds would introduce “more statistical rigour” to the competition.

The experiment is one of several industry benchmarks for AI models. While each model has its own strengths, results from Alpha Arena reinforce the growing view that Chinese AI systems are beginning to rival, and in some cases surpass, those developed by Western companies.

On Stocktwits, the retail sentiment for 'bearish' for BABA, and 'bullish' for OpenAI as of early Tuesday.

For updates and corrections, email newsroom[at]stocktwits[dot]com.<