01What Qwen 3.8 Is and How It Works
Qwen 3.8 is a new AI model from Alibaba that has 2.4 trillion parameters, making it one of the biggest models ever released. A parameter is like a tiny brain cell. The more the model has, the more it can learn and hold.
Alibaba says it is one of the most powerful models on Earth right now, second only to Fable 5.
The key trick is that it does not use all 2.4 trillion parameters at once. Imagine a huge office with 100 experts. When you ask a question, you only call the two or three experts who know that topic.
That is how Qwen 3.8 works. It stays fast and cheaper to run than you might think. That is what makes a model this size possible.
02Qwen 3.8's Massive Parameters and Open-Weight Plan
The biggest news is the size. 2.4 trillion parameters is about 10 times bigger than Alibaba's old flagship model, which had around 235 billion. Most big models before had a few hundred billion. This is a massive jump.
Alibaba also says they plan to open-weight Qwen 3.8 soon. That means they will let people download and run the model themselves. Most top AI models are locked up tight – you can only rent them.
But Alibaba says they will hand this one over. This part is not fully confirmed yet, because their last two big models stayed closed. But if they follow through, it is a big move.
- 2.4 trillion parameters – one of the largest ever.
- Open-weight planned – the public may get the actual model.
- Continuously evolving – Alibaba plans to keep improving it.
03Vision and Multimodal Abilities Beyond Text
One of the lead people on the Qwen team, Shuai Bai, said this is their first trillion parameter model that can see. It can look at images, watch video, and read documents. Then it can act on what it sees.
It is not just a text machine.
You can hand it a picture and it understands it. You can give it a video and it gets what is happening. You can drop a long PDF on it and it reads the whole thing.
This is a huge deal for real businesses.
- Understands images.
- Watches video.
- Reads and processes long PDFs.
- Acts on what it sees.
04How to Get Qwen 3.8 and Try the Preview Today
The preview is live right now, the same day as the announcement. This is different from Alibaba's usual style – they did not make people wait. You can start using Qwen 3.8 today.
- Go to Alibaba's token plan (likely through their cloud platform).
- Sign up for the token plan to access the model.
- Or use Alibaba's coding tools called Coder and Coder work.
- Start with a simple prompt, like asking it to summarize a document.
The move to let people try it the same day builds trust. Talk is cheap, but letting people touch it right away is smart.
05Real Business Uses: Summaries, Content, Support
The episode shows three real ways to use Qwen 3.8 in a small business.
First, you can feed it a folder of screenshots from your tutorials and ask it to write a simple summary for your members. That used to take hours – now it is one prompt.
Second, if you make content, you can feed it your best video and ask it to pull out ten short clips and write captions in minutes. Work that used to eat a whole afternoon.
Third, you can give it a stack of member feedback – messy notes, screenshots, voice memos – and ask it to tell you the top three things your people keep asking for. Then you know what to build next.
- Summarize documents and screenshots.
- Generate content and captions.
- Analyze feedback and find patterns.
None of this needs you to be a coder. No big team required.
06Claims vs. Proof: No Test Scores Yet
Alibaba is making big claims. They say Qwen 3.8 is one of the most powerful models on Earth, second only to Fable 5. But there are no test scores out yet.
No proof on paper. No outside experts have checked it.
Right now, this is Alibaba talking about their own product. Of course they think it is great. The real test comes later when other people run it through the ringer and post results.
Smart people wait for the proof.
The episode warns not to take the claims at face value. They might be totally right, but you should keep a bit of caution. The direction is clear though – models are climbing fast.
A year ago, the biggest were around a trillion parameters. Now a few teams are pushing past two trillion.
07Why Qwen 3.8 Matters for Small Business Owners
Open models like Qwen 3.8 are the key to making AI power accessible to small players. Most top AI models are locked up – you can only rent them. But if Alibaba open-weights this, small business owners could run a top-tier AI on their own.
No gatekeeper.
This matters because the power that used to sit inside giant companies is sliding down to regular business owners. You can build a helper that answers common member questions day and night. You can get your evenings back.
The episode emphasizes that tools like this only help if you know how to use them. That is the gap that holds people back. Joining a community like the AI Profit Boardroom can help you close that gap quickly, with live coaching calls, a 30-day roadmap, and daily tutorials.
08What to Watch For in the Coming Weeks
The preview is live now. The full open release is coming. But the real proof is right around the corner – when independent tests come out.
Keep an eye on the model's performance on standard benchmarks. See if it really beats other top models. Also watch whether Alibaba actually open-weights it, since they have broken promises before.
The episode notes the timing of this release – it came after GLM 5.2 and Kimi K3 shook up the AI world. Alibaba wanted to plant their flag. That is why they dropped Qwen 3.8 at China's biggest AI conference and opened a preview the same day.
For now, you can try it and see for yourself. The model is continuously evolving, so it will keep getting better while you use it.
