Did Qwen 4 Flash beat Opus 5.5? Result from one game-balance benchmark draws 900 likes; its maker says "not correlated with general capabilities"

A screenshot saying an early version of Alibaba's next-generation Qwen 4 beat Anthropic's Claude Opus 5.5 on a benchmark spread on X on Nov. 8.

User Layton Gott wrote, "NO WAY THIS IS REAL," saying Qwen-4-Flash is #1 above Opus 5.5, a 27B model is #4 above Sonnet 5.5, and Flash costs $0.40 a task versus $2.60 for Opus 5.5. The post drew about 900 likes and 90,000 views.

follistatin, who runs the leaderboard, said they had added Qwen-4-Flash-Early (codename "hawk") to GameFeel Bench and noted the score "is not necessarily correlated with general capabilities"; the benchmark evaluates a sense of game balancing and understanding of complex game features. They also fixed a miscalculated audio score.

Alibaba said at its Apsara conference on Sept. 22 that Qwen 4 is in training, and on the 6th posted a teaser with only the line "The becoming begins." It gave no model name, release date or word on open weights. Another user wrote that expectations are running too high, and that since the release keeps slipping, speculation like this cannot be stopped.