Qwen

“The best ai, that I've ever seen. Free video generating, image editing, image generation, coding and other. I use it for making ai content and it is so good at it. I recommend it to all. It actually helps me a lot. ”

“Qwen is dumb, that´s it, no much more to say. The image generation system is even worse, it fails in understaning the most basic commands, make huge mistakes, can t manage images properly, etc. A disaster....”

“einfach nur müll, bilder langweilig und einfach sinnlos zusammen gesetzt. ständig die gleichen bilder egal wie du den prompt veränderst. du hast nen modernen bahnhof und willst zb. eine säule hinzufügen, dann bekommst du eine säule aus dem mittelalter, sinnlos ins bild eingefügt. das macht jedes andere modell besser ”

“As your average user, and after having very good results with ChatGPT for what I use it for, I can honestly say that if Qwen is the future as AI, we are all going to die of stupidity. The images are ugly, slick, boring… the responses are not nuanced. For example, I asked it to draft a specific motion in a specific court for a specific judge, and it started out with Dear Judge WTF. And honestly, if you want to create some visual images just do the hard work and stop consuming this junk. ”

“The best ai, that I've ever seen. Free video generating, image editing, image generation, coding and other. I use it for making ai content and it is so good at it. I recommend it to all. It actually helps me a lot. ”

“Controversial take, but I think there’s a very strong possibility that we’ve seen the last open weights Qwen model. I had assumed the team shakeup after the 3.5 release was related to the original team wanting to make more general purpose models, and the higher ups wanting better agent models. As time has gone on I’ve started to think Alibaba wanted to pivot Qwen away from open weights and into pay-to-play inference. So far that has seemed correct, and the lack of messaging around more open weig”

“Like, look, im not a coder, but i do heavily, and i do mean HEAVILY use ai for design feedback, friend talks, and some homework, and i dont know what this site has, but i swear the rate limits on this thing are non existent, in a good way, Like, i have sent like. hundreds of messages to it daily on end, im not quite sure how many, and i only got rate limited ONCE. and im pretty sure it was some kind of error, and it reset on only 6 hours. When i tried Kimi it did cap in 30 messages.When i tried ”

“Shhhhh just use it. I’m using another “coding” ai like a daily use and it’s BETTER and like you said NON LIMITED. Claude and Kimi are bad and Gemini never limits me but they don’t go in depth in the free tier.”

“Qwen Code comes with a coding agent free tier that feels like unlimited: if one cannot or does not want to pay 20$/mo it's an excellent alternative. I've consumed hundreds of millions of tokens building an app, never got rate limited or kicked out. It's a huge memory hog (I had to tweak the Linux memory management just to get it started). The UI feels much more lagging than Claude , but still better than Gemini which is almost unresponsive on my mid-low tier laptop. The dad jokes i”

“great ai but ruined 2 hours of work and broke real life hardware”

“I'm currently using Qwen3.5 4B as the brain of my local AI assistant because my hardware is relatively limited. One thing I really like about it is the speed. On my system, I'm getting around 40–50 tokens/sec, which makes the interaction feel surprisingly close to real-time. So I don't want to move to a much larger model if that means losing that responsiveness. What I'm wondering is whether there are any newer or alternative models in roughly the same size/memory/compute range that are noticeab”

“everytime I write something it says—Invalid input the last message must be a user message or tool message. what's about 'unknown error'?! it always shows that! I am fed up! Hey, it has daily chat time limit. I don't like that, please fix it soon!”

“Seeing Alibaba Cloud in the same conversation as AWS and Azure is wild—but after this walkthrough, it actually makes sense. Model Studio’s no-code setup and Qwen’s capabilities look seriously practical for anyone building AI tools. Great breakdown, super easy to follow!”

“wait .. I thought to run Qwen 3.8 Max locally—even heavily compressed (quantized to 4-bit)—you need roughly 1,300 GB to 1,500 GB of VRAM.”

“But it needs to be usable locally not in china cloud, The models need to be smaller and smarter not the other way round. This is just stupid.”

“3:36 Let me just pull out my 2 TB of vram so I can run this locally...”