Qwen turned the tide with a 25% ROI to follow Grok closely, while Claude went from profit to loss.

BlockBeats News, October 22nd, according to on-chain AI analysis tool CoinBob (@CoinbobAI_bot), in the mainstream AI large model trading competition, the initial capital of the six major AI models was all $10,000. Claude dropped from second place on the leaderboard yesterday to fourth place, with its initial capital turning from a profit to a loss of 13%, while Qwen achieved a comeback with a return rate of over 25% last night, approaching GROK. The specific performances are as follows:

Qwen 3 closed its BTC long position last night at a range of $110,200 to $112,000 at 22:00, realizing a profit of $2,336. Afterward, it experienced some small losses due to frequent opening and closing trades in BTC and ETH during pullbacks and surges. The current total account value is about $10,300, with a short position in SOL holding a 13% unrealized gain. Claude 4.5, on the other hand, opened new long positions in ETH, BTC, and SOL at $111,800 for BTC after 1 a.m. last night, but was forced to liquidate the positions for a loss before 7 a.m., losing nearly $3,000. The current total account value is about $8,737, with long positions in XRP (31% unrealized gain) and DOGE (6% unrealized gain).

Earlier reports indicated that the innovation team nof1ai conducted a practical test: providing $10,000 in real funds to 6 mainstream AI models—Deepseek V3.1, CLAUDE, Grok 4, Qwen 3 Max, ChatGPT, and GEMINI—and allowing them to autonomously trade BTC, ETH, SOL, and other perpetual contracts on Hyperliquid.