Qwen returned to profitability with a 25% return rate, chasing Grok, while Claude went from profit to loss.

On October 22, according to the on-chain AI analysis tool CoinBob (@CoinbobAI_bot)), in the mainstream AI model trading competition, the six major AI models started with an initial capital of $10,000. Claude dropped from second place to fourth yesterday, turning from profits to losses of 13% compared to its initial capital, while Qwen turned from losses to profits with a return rate of over 25% last night, approaching GROK. The specific performance is as follows: Qwen 3 escaped the peak for take profit on a long order for BTC in the range of $110,200 to $112,000 last night at 10 PM, achieving a profit of $2,336, and later incurred small losses by chasing the price on BTC and ETH during the pullback. The current total value of the account is about $10,300, with existing holdings of a short order for SOL, unrealized gains of 13%. Claude 4.5 opened new long orders for ETH, BTC, and SOL at $111,800 after 1 AM last night, but was forced to close positions for stop loss before 7 AM, realizing a loss of nearly $3,000. The current total value of the account is about $8,737, with existing holdings of a long order for XRP (unrealized gains of 31%) and a long order for DOGE (unrealized gains of 6%). Previously reported, the innovation team nof1ai conducted a practice: providing $10,000 of real funds for six mainstream AI models - Deepseek V3.1, CLAUDE, Grok 4, Qwen 3 Max, ChatGPT, and CEX - allowing them to trade BTC, ETH, SOL, and other perpetual futures autonomously on Hyperliquid.