Qwen3.8 Max Sits Second, Not First, on the Key Chinese AI Leaderboard

The clearest number in the room is Qwen3.8 Max's score of 73.2, good for second place on the BenchAlign v5 ranking. That is a strong showing, but it sits behind Moonshot AI's Kimi K3, which topped the same board at 74.4. On a separate industry tracker, the gap widens further: the highest-placed Chinese model there is Z.ai's GLM-5.3 at 45, seventh overall, with Alibaba nowhere near the front of that list either. Alibaba's marketing has been loud. The company touts Qwen3.8-Omni-Flash, a million-token, four-modality system, and claims audio costs down more than 98% and a 25% average score jump across 29 internal tests. Those are company statements, not independent rankings, and they describe efficiency and multimodal reach rather than proven leadership on the text benchmarks that actually decide this question. Where Alibaba does lead is narrower turf, topping a coding-specific arena board ahead of Kimi K3 Max. That is real, but it is not the same contest as the overall text leaderboard, where Kimi K3 and GLM-5.3 both outrank Qwen3.8 Max as of mid-September. Put simply, two independent trackers agree Alibaba is not first on the measure that matters, and its own promotional claims about the Omni line have not yet been verified against rivals. The one thing worth watching before month's end: whether an independent leaderboard update actually vaults Qwen past Kimi K3 or GLM-5.3, because right now the paper trail says it hasn't.