I slightly disagree with you. First of all, Qwen 3.6 Plus was released on March 30th, and the open-source model came out 20 days after that. Now that 3.7 Max is out, waiting about a month for the open-source version to drop is totally normal. Regarding Smaller vs. Larger Models First, imagine a 9B model came out at the beginning of March. Not even a month has passed between releases, and you're suggesting they put out a model that’s only 10% better than the previous one? What’s the point? Just so people like you can write, "Why did you even release this model? It's basically just like 3.5." There is zero sense in bothering to release a model that will be irrelevant from day one and only marginally better. In reality, they’d release a model only to have it get completely trashed, and on top of that, nobody would even use it. Why would they want that kind of bad PR? The Strategy for Big Models Next, regarding the large models, my assumption is pretty much the same. They revolutionized the game with the 27B and 35B models. Either they haven't gathered enough data yet to release a local 120B, 350B, or something similar, or they are holding back. What will you say if they release, say, Qwen 4 220B, which goes head-to-head with GPT-5.5 and Opus 4.6? You'll be kissing their feet! But if they release a Qwen 3.7 220B model right now that's only 10–20% better than their Qwen 3 model? You'll tear them apart and say you're never using their stuff again. I hope I've made my point clear.