Qwen 3.7 Preview

Alibaba’s Qwen 3.7 Preview just strutted onto the public AI leaderboards and, on paper, it’s a solid flex: #13 overall for text and #16 for vision, helping Alibaba claim the #6 lab spot in text and a strong showing in image understanding too. But if you thought the crowd would simply clap and move on, absolutely not. The real action is in the replies, where the mood is a chaotic mix of hype, impatience, and nitpicking.

One camp is excited that Alibaba keeps shipping and teasing more to come, with one commenter linking the company’s own announcement like they were dropping celebrity receipts. Another group immediately turned the moment into a leaderboard soap opera: “Where’s Grok 4.3 on the leaderboard?” Translation: nice entrance, but the audience is already asking who’s missing from the party. Then came the painfully relatable hardware crowd, with users basically saying, cool, but can my laptop survive this? One person was waiting for a smaller Qwen model they could actually run on a Mac, only for Alibaba to drop a bigger new preview instead.

And then there’s the brutally honest performance discourse. One fan praised an earlier Qwen model as the first one that doesn’t constantly “lose its mind,” while still admitting it sometimes falls into weird loops. Ouch. Another commenter defended vision features as wildly underrated, while also dragging Qwen’s familiar writing style and wondering why Google’s Gemma sounds smoother. The biggest underlying tension? People are impressed by the rankings, but they’re still begging Alibaba to keep its models open, useful, and less bizarre in everyday use.