Claude Code Is Dead? China Is Giving AI Away for FREE (Qwen + OpenCode)
A 14-minute walkthrough of wiring a free Qwen model from Iteracompute into OpenCode, then building a working CRM without spending a dollar on AI tokens.
Posted
today
Duration
Format
Tutorial
educational
Views
15K
340 likes
57 · 43
Big Idea
The argument in one line.
A free coding agent (OpenCode) plus a free Qwen model from Iteracompute can build a working app from one plain-English prompt, with zero AI token cost for personal projects.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You want to build software by describing it in plain English but don't want another monthly AI subscription.
You're curious whether Alibaba's Qwen models can stand in for paid coding assistants on side projects and prototypes.
You've never set up an AI coding agent and want a click-by-click walkthrough of installing one and adding an API key.
SKIP IF…
You need to work with client data, passwords or private files, since the video itself warns free access is not private.
You want a rigorous benchmark against Claude Code, because the only evidence offered is a single CRM demo and one Alibaba claim.
TL;DR
The full version, fast.
Maitis argues you can stop paying for AI coding tools by pairing OpenCode, a free desktop coding agent, with a free Qwen model from Iteracompute, which offers up to 30 million free tokens a month. He installs OpenCode, creates an Iteracompute API key, adds it as a provider, and sends one detailed prompt that produces a working CRM with customers, deals, tasks and reports. The usage dashboard shows zero cost. Caveats: only the free Qwen 3.8 27b model is free, free access can be slow, nothing private should go through it, and the offer runs until November 24.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Frames Qwen as a free alternative, poses a yes-or-no challenge, and previews three levels.
01:42 – 02:32
02 · Midas.ai plug
Pitches the companion step-by-step guides on Midas.ai and asks for a subscribe.
02:32 – 03:56
03 · Level 1: install OpenCode
Downloads OpenCode, launches it and opens the Providers preferences.
03:56 – 05:45
04 · Level 2: get free Qwen tokens
Claims the Iteracompute offer, creates an API key and adds the provider to OpenCode.
05:45 – 07:06
05 · Testing the connection
Sends a test message, shows the paid-model low-balance error and explains the empty-reply fix.
07:06 – 08:46
06 · Level 3: prompting the CRM build
Sends one detailed prompt and watches OpenCode plan and generate the project.
08:46 – 10:40
07 · Testing the finished CRM
Edits customers, moves a deal to won and completes tasks, showing changes flow into reports.
10:40 – 11:30
08 · Cost check
The Iteracompute usage dashboard shows zero dollars spent on tokens.
11:30 – 12:37
09 · Bonus: free backup models and Cline
Switches to bundled free models when busy and reuses the same provider in Cline.
12:37 – 14:23
10 · Caveats and wrap-up
Offer ends November 24, free can be slow and is not private, then recap and outro.
Atomic Insights
Lines worth screenshotting.
Up to 30 million free Qwen tokens a month is enough to build a full multi-page CRM from a single prompt without a paid balance.
Not every model in an Iteracompute list is free: selecting a bigger Qwen model returns a low-balance error, so check you are on the free one.
An empty reply from a reasoning model usually means the token cap was spent on thinking, so raise max tokens or turn thinking off.
An API key is just a password that lets one app talk to another, and it is shown once, so copy it immediately.
A custom provider needs only four things: an ID, a name, the base URL from the API docs, and a key plus model ID.
A single detailed prompt that lists entities, pipeline stages and an instruction to leave nothing half-finished beats iterative vague prompting for a first build.
A demo is only convincing if edits propagate: moving a $50,000 deal to won changed revenue, open pipeline and the dashboard.
Free tiers trade money for privacy and speed, so keep client work, passwords and private files out of them.
Free promotional model access has an expiry (November 24 here), and continued free access is not guaranteed.
The same provider credentials work in a second agent (Cline), so one free allowance can power multiple coding tools.
Free models bundled into the agent itself make a fallback when the main provider is busy.
Takeaway
Free AI coding works, within limits
WHAT TO LEARN
A free coding agent and a free Qwen model can build a working app from one detailed prompt, as long as you accept slower speed, no privacy and an expiring offer.
An AI coding agent and the model behind it are separate pieces, and you connect them by adding a provider with a base URL and an API key.
Check that the exact model you select is the free one, because paid models in the same provider list fail with a low-balance error.
If a reasoning model returns an empty reply, raise the max token setting or turn off thinking before assuming the setup is broken.
Give the first build prompt a full spec: entities, pipeline stages, reports, and an instruction not to leave anything half-finished.
Judge a generated app by whether actions propagate, such as a moved deal changing revenue and pipeline figures on the dashboard.
Keep client work, passwords and private files out of free tiers, and treat promotional free access as temporary.
Glossary
Terms worth knowing.
OpenCode
A desktop AI coding agent that takes plain-English instructions, writes code, fixes errors and assembles a project on your computer.
Provider
The company or service that supplies the AI model an agent uses to think. The agent is the interface, the provider is the brain.
API key
A secret string that lets one app call another service on your behalf. It is usually shown once and should not be shared.
Base URL
The web address an app connects to when it talks to an AI provider's API.
Token
The unit of text an AI model reads and writes, and the unit providers meter and bill by.
Qwen
A family of AI models from Alibaba that can write code and solve problems.
CRM
Customer relationship management software that tracks customers, deals, tasks and revenue in one place.
Cline
Another AI coding agent (transcribed as Klein) that supports bring-your-own-key providers.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
17px
For years, the most powerful AI models came from OpenAI, Google, and Anthropic, and you had to pay every month to build with them. Then, AI tools from China started showing up, and they began going head to head with all of these big platforms. But some of the most powerful ones can do a lot of the same work, and most people still don't know about them.
So the big question is, are they really free? And what can we actually build with them? So what if you could stop paying for other people's software and build your own?
What if you could do it without writing a single line of code, just by telling an AI what you want in plain English? And what if that AI didn't cost you a single dollar? That's exactly what we're going to find out today.
There's a free AI model from China called Quen. It comes from Alibaba. And with the right free workflow, it can do a lot of what Cloud Code does.
It can write code, solve problems, and help you build real projects in plain English. In Alibaba's own coding tests, it even beat an older Claude Opus model. But when most people hear that, they think they'll need a really powerful computer to use it.
You don't. There's a way to get up to 30 million free Quen tokens a month without paying a single dollar. So today, I'll show you exactly how to get those free tokens.
And then we'll put them to the real test. Can a free AI model really build a full platform, the kind of software companies pay for every month, without us spending a single dollar? Drop your answer in the comments before we start.
Yes or no? We'll do this in three levels. In level one, we'll set up a free AI developer on your computer.
In level two, we'll get our free Quen tokens and connect them to it. And in level 3, we'll use those free tokens to build our platform from scratch. And stay with me till the end, because I'll show you the finished platform, exactly what it cost us, and a bonus that can save you when the free model gets busy.
But before we jump in, give me 10 seconds. One missed command can break this whole setup, and that's exactly why I built Midas .ai. Every video I make turns into a free step -by -step guide there, with the commands, prompts, and links ready to copy and paste.
And today's setup is waiting for you too. The link's in the description. And if you're new here, hit subscribe, because we break down free AI stuff like this every day.
Alright, let's get started.
Level 1. Installing OpenCode Before we build anything, we need someone to do the actual building. And we're not hiring a developer, we're using one.
It's called OpenCode. Think of it as a developer that lives on your computer. You tell it in plain English what you want, and it writes the code, fixes the errors, and puts everything together for you.
So, let's get it set up! Search for OpenCode on Google and open the official website. Right on the homepage, you can watch it break a job into smaller tasks and work through them.
Now, click Download, and you'll see options for different computers. I'm on a Mac with Apple Silicon, so I'll choose that one. If you're on Windows, just pick the Windows version.
Once the file is downloaded, open it, drag OpenCode into your Applications folder, and launch it. And just like that, we're inside! Let's take a quick look around.
This is where we'll chat with our developer, and this menu is where we choose which AI model does the thinking. You'll notice a few free models here already. Hold on to that, because it's going to matter at the end of this video.
Next, let's open Preferences and go to Providers. Think of a provider as the company that supplies the AI brain to our developer. Right now, we haven't connected the one we need.
So here's what we've got so far. OpenCode is our developer, and it's ready to work. But there's still one piece missing.
Our developer needs a powerful brain, and we don't want to pay for one. That's exactly what we're fixing next. Level two, getting 30 million free tokens.
So, now we've got our developer. But where do those free Quen tokens actually come from? That's where Iteracompute comes in.
Think of it as the place that gives us access to Quen, along with a free monthly limit. Open iteracompute .com. Right on the page, you'll see the offer.
Up to 30 million free tokens a month. So, let's go and claim it. Iteracompute even has a page that explains how to use free Quen with open code.
So, let's click Get a PI Key. Here, you can see the free limits again. I'll click Continue with Google and choose my account.
And there we go! We're inside the dashboard and our usage is sitting at zero. Let's take a quick look around.
Models shows everything Iteracompute offers, and the Free section shows the limits we get. Then, in the API docs, you'll find the base URL, which is just the address our app connects to. Hold on to these pages, because they can save you in a second.
Now, here's the important part. Open API Keys and click Create API Key. An API key sounds technical, but it's really just a password that lets one app talk to another.
I'll name mine Test Key. Create the key and copy it right away because you'll only see it once. Save it somewhere safe and don't share it with anyone.
Then click I have stored it and our new key shows up in the list. Now, let's head back to OpenCode. Open Preferences, go to Providers, click show more providers and find itera compute click it paste your key and continue next choose the models you want and confirm and that's it our developer is now connected to its new brain if you don't see itera compute in that list don't worry you can add it yourself create a custom provider Enter itiracompute as the ID and itiracompute as the name, then paste the base URL from the API docs, your API key, and the model ID from the Models page.
Click Submit, and you'll end up in exactly the same place. But before we move ahead, let's quickly check whether it's actually working. Back in the chat window, you can set how hard the model thinks, and this menu shows the itiracompute models we just connected.
I'll pick one, send a simple, hi, and there we go. We got a reply, so our connection is working. Pretty cool, right?
Now, be careful here, because not every model in this list is free. When I switch to this bigger Quen model, I get an error saying my balance is too low for a paid model. So, make sure you stay on Quen 3 .827b, because that's our free model.
And if you get an empty reply, don't panic. Raise the max token setting or turn off thinking and try again.
So let's quickly look at everything we've done so far. We've installed OpenCode as our developer, created our free key, and connected Quen as its brain, all without paying for an AI subscription. And if you've made it this far with your setup working, quick favor, hit subscribe, because we break down free AI tools like this every single day.
But getting free tokens is the easy part. Now we need to find out what they can actually build. Level three, building our platform.
Now comes the real test. We've got our free AI developer, we've connected Quen, and everything is ready. But can this setup actually build something people normally pay for?
Well, let's find out. For our challenge, we're building a CRM. If you haven't used one before, think of it as a control room for a sales team.
It helps businesses manage customers, deals, tasks, and revenue in one place. And companies pay for software like this every single month. So let's see how close our free setup can get.
Head back to OpenCode and make sure Iteracompute and Quen 3 .827b are selected. Now, we'll give it one detailed prompt asking for customers, sales reps, deals across six stages, tasks, revenue reports, and notifications. And we're also telling it not to leave anything half -finished.
I'll leave the full prompt in the description, so you can try it yourself. Let's send it and see what happens. OpenCode starts reading our workspace and planning the project.
Your computer might ask for permission to access a folder, so make sure it's the right one before clicking Allow. On my computer, it even found two older copies of the project and asked what to do, so I chose a fresh folder to start from scratch. Now our AI developer gets to work.
It creates the project files, works on the design, and starts putting everything together. And the interesting part? We're not writing a single line of code ourselves.
Quen is doing the work through OpenCode, using those free tokens we just set up. After a while, the build finishes. So let's open our CRM and see what we've got.
And look at this! We've got a dashboard showing our total revenue, open deals, customers, and pending tasks. There are even charts showing sales performance and recent activity.
But a good -looking dashboard is one thing. Can we actually use it? Let's open customers.
Here, we can see our customer records, their companies, and assigned sales reps. We can even filter them by status. Now, I'll open one customer, change the assigned sales rep, and save it.
And there we go. The update is confirmed. Next, let's check the sales team.
We can see our reps, their assigned accounts, and the deals they're handling. Now, let's move to the deals board. We've got six stages, from new leads all the way to won or lost.
I'll take this $50 ,000 deal and move it from negotiation to won. And just like that, the deal is updated. We can even search for it to check the changes.
And because that deal has moved, let's see what happens to our reports. Here, we can see our revenue, sales performance, and monthly targets, all in one place. But we're not done yet.
Under Tasks, we can see what's open, overdue, and completed. Let's mark one overdue task as finished. And there it is, right in our completed list.
We've also got a notification panel showing recent updates and deadlines. And inside Settings, we can even switch between light and dark mode. Now, let's head back to the dashboard.
Notice something? Our revenue has increased, the open pipeline has changed, and we've got one less pending task. So this isn't just a good -looking interface.
We can actually edit customers, move deals, complete tasks, and see those changes reflected across the CRM. And all of this came from the prompt we gave Quen. Now, remember the question I asked at the beginning?
Can we build something people normally pay for without spending a single dollar on AI tokens? Well, let's check. Head back to Iteracompute and open the usage dashboard.
Here, we can see how many requests we've made and how many tokens we've used. And the total AI token cost of building this CRM? Zero dollars.
We started with a free coding agent, connected a free QEN model, and built an interactive CRM without paying for AI tokens. So if you know someone who's still paying for AI coding tools, or someone who's been putting off an idea because of the cost, share this video with them. It could help them get started without another monthly bill.
And we've proved that the free setup can build something useful. But I've still got one more thing to show you. Because there's another way to get even more out of these free AI tools.
Let's get into the bonus. Remember the free models I told you to hold on to in level 1? Well, here's your bonus.
When a Terra Compute gets busy, you don't always have to sit and wait. Just open the model menu in OpenCode, look for the models marked free, and pick one. And just like that, you've got a backup brain ready to go.
These models can change over time, so choose whichever free options are available. But hold on, I've got one more surprise for you. What if you wanted to use those same free Quen tokens outside OpenCode?
Well, you can actually do that too, using another AI coding agent called Klein. And the best part? You don't need another API key or a separate account.
We can use the same Iteracompute setup we created earlier. So, search for Klein, open the official website, and download it for your computer. Once it's installed, open Settings, go to Providers, and select Bring Your Own Key.
Search for Atera Compute and choose it from the list. Now, paste the API key we created earlier. The base URL should already be there, so just check that it's correct.
Once the provider is configured, head back to your session, switch to Atera Compute, and select the free QEN 3 .827b model. Let's send a quick hi to test it. And there we go!
We've got a response from Quen, this time inside Klein. So now we've got free backup models in OpenCode, and another way to use our Quen tokens through Klein. Two coding agents, one free API allowance, and no extra AI subscription required.
Pretty useful, right? Now, I do want to be honest about a few things. Iteracompute says this free Quen offer runs until November 24th, and free access after that isn't guaranteed.
So if you want to try it, do it soon and as you just saw free can also be slow so keep client work passwords and private files out of this setup because free doesn't always mean private but for personal projects and testing your ideas this is a great place to start and that's all three levels we started by installing open code got our free quen tokens from itera compute and connected everything together Then we put the entire setup to the test by building a real CRM with $0 spent on AI.
For years, powerful AI meant paying every month. Today, you can start with one free key and a good idea. And all the prompts and links from today's video are in the description too.
If you want the complete step -by -step guide, it's waiting for you on midas .ai. And if you want to stop just hearing about AI and actually start building things with it, subscribe to Midas. We're constantly testing free AI tools, finding alternatives to expensive software, ensuring practical workflows you can use yourself.
Now, remember the question I asked at the start? Tell me in the comments if your answer was right, and what you'd build with 30 million free tokens. Thanks for watching, and I'll see you in the next one.
The Hook
The bait, then the rug-pull.
The title declares Claude Code dead, but the video is a practical three-level setup guide: install OpenCode, claim free Qwen tokens, and test the result by building a CRM at zero AI cost.
Frameworks
Named ideas worth stealing.
01:21list
Three levels to a free AI builder
Install a free coding agent (OpenCode)
Get free Qwen tokens and connect them
Build a real project with those tokens
The video's structure: agent first, brain second, proof third.
Steal forany setup tutorial
CTA Breakdown
How they asked for the click.
VERBAL ASK
13:53subscribe
“If you want to stop just hearing about AI and actually start building things with it, subscribe.”
Repeated subscribe asks throughout, plus a pitch for the Midas.ai companion guide.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
A three-level walkthrough that wires a desktop coding agent to a reverse-engineering toolkit and free OpenRouter models, then builds Slack and ChatGPT lookalikes with no paid API.
A former WordPress builder walks through the 18 objections he hears about shipping client marketing sites with Astro, Claude and Cloudflare instead of a page builder.