Qwen 3.7 Max BEATS Opus 4.7???

Added: 2026-05-22

[00:00:00]So, there's a new model that has just hit OpenRouter, Qwen 3.7 Max. This is something that I'm extremely excited for. However, just so everybody knows, this model is expensive. It's cheaper than other flagship models like Opus GPT-5.5, that kind of stuff, but it is actually pretty expensive for a Chinese model. So, I don't want people to think that this is like a cheap model or anything like that. But, this will be very, very interesting to see how good this model actually is. So, Qwen 3.7 Max is the flagship model in Alibaba's Qwen 3.7 series. This is brand new, 3.7. It supports text input and output and is a designed for agentic-centric workflows with particular strengths in coding, office, and productivity tasks and long horizon on terminal execution.

[00:00:48]That's Open Core.

[00:00:50]The model offers noticeable gains in coding and agentic performance. So, I'm actually just testing this right now, but unfortunately, I've run out of Open Router credits because this model is so damn expensive.

[00:01:03]Guys, I have to say, this is an extremely expensive model. I don't actually know what happened here, but I topped up $16, which is normally enough to do an entire build, and it barely even got through part of it. So, I mean, this is going to end up being a $30 to $40 build. So, I'm not sure if this is actually worth it for people, but I still do want to test out this model.

[00:01:22]So, I think something is going on in the background here. First of all, holy crap, that is quick.

[00:01:31]Oh, no, sorry. I was looking at the wrong one. It is It is really fast, though. It seems faster than this.

[00:01:35]But, also, it seems to be draining more tokens than it should be. I'm not sure what is going on here.

[00:01:40]Um, but my credits are going down very, very quickly.

[00:01:44]Like, really quickly. Quicker than I would have expected.

[00:01:48]Okay, so this is what I've spent 40 bucks on just for this video. I'm not It's pretty good. I'm not massively impressed by the design necessarily, but it's followed the prompt extremely well and it looks like it's done a very very good job of the technical build. It is pretty much done. I am probably going to just cancel the rest because a lot of it's just, you know, doing the final bits.

[00:02:08]But I mean, yeah, it's done a pretty damn good build, I have to say, but that was $40 I just spent on on this. So, I'm actually probably going to make a loss on this video. So, yeah, about that. Rip. Um But yeah, I mean, the quality's good, the font choice is good, the colors are good. There's no weird colors on colors.

[00:02:27]Um it's a good design, generally speaking, I would say. It's done the It's followed the prompt very very well. Is this an open-source model? Uh Qwen 3.7 Hugging Face. Let's have a look.

[00:02:42]Uh doesn't look like it is yet. Not sure what they're planning on doing with that. Let's just see. So, this is Qwen 3.7 Max, the Agent Frontier.

[00:02:54]Uh this is 1 day ago. I didn't even see this model, apparently. Uh I just completely missed it. Uh I really need to get inside Alibaba Cloud Model Studio. I would love to be inside there, but they've got this weird thing. Um let's just have a look at some of the benchmarks. So, yeah, claiming to be better than Opus 4.7, standard procedure. It does look pretty good though, I have to say. And it is actually cheaper than Opus 4.7 by a lot.

[00:03:17]It does seem to be wasting a lot of credits though, so I do have to um look into that. It looks like it's just calling, you know, things for no reason or whatever.

[00:03:29]With all that being said, guys, this is looking like a very very good model from China, probably the best Chinese model right now, but I mean, the usage limits don't make any sense. Once it's available inside Kyoda or Coda, or whatever it's called, then I'll probably do another test inside Coda, cuz I do like Coda. I do pay for Coda, last time I checked, but maybe I'm not paying for it right now. And it normally gives you a little bit more usage. So, I will also test it in site coder. But with that being said guys, just a quick word from our sponsor, me. This is harborseo.ai.

[00:04:00]It is a content generator which I've been working on for the last few months.

[00:04:04]And the results kind of speak for themselves. 600 pages published, 350,000 impressions, 3,000 clicks. I'll probably make a loss on this video guys spending 40 bucks on doing the test. So, please go and support me by checking out harborseo.ai. I would really appreciate it. This is a good model, but I don't think it's worth the price to be honest with you.

[00:04:26]Now, if you're inside Alibaba's coding thing, then it might be worth it. But there's something going on with open router. It's making it's making way too many calls like 902 902 902 902 901 901.

[00:04:39]It's making so many different calls and it seems to be using a lot of input as well. Obviously, the input kind of stacks up, right? Um we do know that already. But yeah, I mean it seems like a decent model, but I'm I'm not really sure if it's worth the price. With that being said guys, there's a link for harbor in the description and in the pin comment. You can go get yourselves a free trial of the best quality and best priced AI SEO content generator on the market. Thank you so much for watching.

[00:05:06]If you are watching all the way to the end of the video, you're a absolute legend. I'll see you very very soon with some more content.

[00:05:11]Peace out.

[00:05:12]I'll just say now guys, the build did actually finish. Um and yeah, I mean it it's done a good job. I have to say it's done a very very good job. This is like Opus 4.6 levels, I would say, if not slightly better. Thanks guys and peace out.