Original author: 1912212.eth, Foresight News

The crypto community's buzzing NOF1 AI Trading Competition (Alpha Arena) is in full swing. Hosted by nof1.ai, the competition launched on October 17, 2025, and will run until November 3. Six top AI models—DeepSeek Chat V3.1, xAI's Grok 4, Anthropic's Claude Sonnet 4.5, Alibaba's Qwen3 Max, Google's Gemini 2.5 Pro, and OpenAI's GPT-5—each received $10,000 in starting funding to trade BTC, ETH, SOL, XRP, DOGE, and BNB perpetual contracts on Hyperliquid.

After the competition officially began, the winners of the various large models were unclear, with the price fluctuating primarily around $10,000. However, on October 19th, a clear watershed emerged between the competitors. On that day, the price of Bitcoin rose from $106,000 to $109,000. Significant increases followed on October 20th and 21st.

As of 11:00 AM on October 22, DeepSeek has consistently performed well, ranking first with a value of $11,129. Its strategy uses 10-15x leverage to go all long.

Claude was quickly left behind by the evening of October 21. During this timeframe, BTC rose from its bottom to $114,000, but then quickly fell back to $108,000.

GPT and Gemini, two struggling peers, have lagged far behind, falling significantly behind their competitors since October 19th. GPT's performance has been particularly disastrous, with only $3,578 remaining, while Gemini's is at $4,408. Its strategy is the exact opposite of DeepSeek's: shorting all cryptocurrencies.

On October 23, the market suddenly changed, BTC rebounded to $110,000, and Qwen, which had been unknown until now, surpassed DeepSeek by a large margin and ranked first in profitability.

Some readers may be curious about how it managed to quickly take the lead. The answer is simple: it only trades BTC, sees the right opportunity, and then takes a large position.

This advantage was further expanded as Bitcoin surged all the way to $115,600 on the morning of October 27.

However, as of now, Qwen's position has been adjusted to 25 times long on ETH, resulting in losses and falling from first to second place.

DeepSeek seized the opportunity to reclaim first place from second place. Its strategy employed a comprehensive 10x long/short position across six cryptocurrencies, with its largest long position adjusted to ETH. BTC was its second-largest position.

Grok and Claude competed for third and fourth place respectively, with decent performances.

GPT and Gemini are firmly in the last two places.

The current value of GPT's position is $4,200, while Gemini's is $4,400. Polymarket data shows that the market currently has a 50% probability of betting on OpenAI as the biggest loser, while Google has a 45% probability.

At present, all contestants in the AI big model competition have chosen to go long across the board. Perhaps the pessimism and panic caused by the "1011 crash" have quietly dissipated.

Alpha Arena officials revealed significant differences in investment decision-making among different AI models. DeepSeek tends to employ quantitative analysis, while Grok excels at flexibly adjusting strategies based on market conditions. While all models receive the same market data input, their differing training data and reasoning methods often result in significantly different investment recommendations.

GPT was relatively conservative and missed the initial market rebound. Although it later learned to switch to a long-term strategy, it was already in a losing position. Because the decision-making chain is relatively long, the longer it is, the more likely it is to amplify bias, which is a significant factor in its decision-making errors and large losses. Gemini's trading frequency is very high, which is a taboo in the trading world because high-frequency trading amplifies decision errors, which means that decisions must be extremely accurate.

The outcome is still unknown. On November 3rd, we will see who will take first place.