FREE Mode Coder: This is LITERALLY $0 - NO SUBSCRIPTION FREE CODER!

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: FREE Mode Coder: This is LITERALLY $0 - NO SUBSCRIPTION FREE CODER! · Channel: AICodeKing · 09:37 · Watch on YouTube

Topics: AI agents, Software engineering with AI, AI developer tools, Markets and economy

Summary

The video demonstrates Verdant’s free mode for AI-assisted coding and explains how to use it in the desktop app. In a prepared JavaScript shopping cart project, the agent reproduced four failing tests, fixed the calculation, and passed all five tests without changing the test file. The presenter explains that free mode has an allowance that refreshes every five hours as well as a weekly cap, and contrasts it with eco and prime. They recommend starting with focused, verifiable tasks and inspecting the agent’s changes and results.

Key points

Verdant’s free mode can be selected in the desktop app when creating a regular project task.
The demo asks the agent to fix a JavaScript cart total that mishandles string inputs and ignores item quantities.
The agent ran the existing tests before editing, reproduced four failures, made a two-line change, and then passed all five tests.
The fix converts prices and quantities to numbers, multiplies them, and rounds the final total to cents; the test file remains unchanged.
Free mode refreshes every five hours and also has a weekly cap; the announcement does not specify a fixed number of messages.
Free is available without a subscription; eco uses zero credits but requires a subscription, while prime spends credits for access to stronger models.
Suggested tasks to try include tracing unfamiliar code, making a small interface change, and drafting documentation.
The presenter recommends defining how to verify a result before sending a request, then inspecting the changes and checking the outcome.

Chapters

Start	Chapter	Summary
00:02	Introducing Verdant free mode	The presenter introduces free mode, outlines the demo, and explains how Verdant’s project-based coding workflow works.
01:07	Setup and shopping cart test	The video shows how to select free mode and describes the cart bug, the expected behavior, and the tests.
03:13	Agent fix and test results	The agent reproduces the failures, makes a small change, and passes all five tests without editing the test file.
05:19	Usage limits and mode differences	The presenter explains the five-hour refresh, weekly cap, and differences between free, eco, and prime.
06:55	Other tasks and final advice	The video suggests other focused tasks to try and emphasizes checking results, learning from changes, and evaluating usefulness.

Quotes

Verdant has opened free mode to everyone. — 00:02

Five passed, zero failed. — 04:17

free mode refreshes every 5 hours and also has a weekly cap. — 05:19

Free is available without subscribing. — 06:24

Try a task you’d otherwise handle yourself, inspect the output, and see whether the time saved is meaningful for your workflow. — 08:32

Concepts and entities

Verdant, free, eco, Prime, JavaScript, testes

Full transcript (original language)

00:02 [music] >> Hi. Welcome to another video. So, Verdant has opened free mode to everyone. And this is a pretty interesting update if you want an AI coding agent without adding another subscription to your monthly expenses. You can open the desktop app, choose free mode, and start working without buying a subscription or credits. There is an allowance that refreshes, and there is also a weekly cap, which I’ll explain in a bit. But the part I wanted to check was what you can actually do with it, because having a free option is useful when it can get through a task you care about. So, in this video, I’ll

00:36 show you how to start using it, walk through a small coding test I ran in the desktop app, and explain how free is different from eco and prime. I’ll also talk about the kinds of tasks I would try first if I were using this on a limited budget. So, let’s get right into it. If you haven’t used Verdant before, it’s an AI coding tool where you open a project and describe what you want done. The agent can work with the files in that project, make changes, and run commands to check the result. For example, you can give it a bug to investigate and ask it to reproduce the problem before making a change. You can then look at the files it edited and the tests it ran, all as part of the same

01:07 task. That is the workflow we’re going to use here. I want to see whether it can read a small piece of code, understand why it is broken, fix it, and verify the fix. Now, the 2.13 update also changed how you choose the AI for a task. You have free, eco, and prime, and Verdant can route the task automatically based on the mode you select. Personally, I like having this option. When you’re just trying to fix something small, choosing between a bunch of model names can become another decision before you’ve even started working. Here, we

01:38 can select free and get on with the task. In the desktop app, the description says basic models for light work, which gives you a reasonable idea of where to start. Now, let me show you the setup. First, download Verdant from its official website, install the desktop app, and sign in. Once you’re inside, go to projects and click new task. You’ll see the option to select or create a project. For this example, I’m choosing an existing folder because I’ve already prepared a small project with the code and its tests.

02:09 Then, beside the message box, open the mode selector and choose free. You can see that it is selected here. Now, you can type your request and send it. One detail to keep in mind is that I’m using a regular project task. Manager currently supports eco and prime. So, if you’re following this particular free mode walk-through, start from projects and new task as I’m going here. Anyway, let’s come to the example. I have a tiny JavaScript shopping cart function. Its job is to take some items, multiply their prices by their quantities, and return the total. The

02:41 problem is that values coming from a form can arrive as text. So, a price that looks like $12.50 might actually be stored as a string, and the original function just adds that value to the running total. With strings, JavaScript can end up joining the text together. The function also ignores the quantity, which means buying three of something doesn’t produce the correct total, either. This gives us a straightforward problem with a result we can check. If one item costs $12.50 and I buy two, the total should be $25. I’ve also included five tests. They cover values coming in as strings,

03:13 multiple items, ordinary numeric values, a small decimal calculation, and an empty cart. Before the fix, four of those tests fail. Only the empty cart test passes because the function starts with zero and has nothing to add. So, here’s what I’m asking Verdin to do. Fix the cart total calculation. Prices and quantities may arrive as strings from a form. Return a number rounded to cents, and return zero for an empty cart. Run the existing tests first, make the smallest fix, and run them again. Leave the test file unchanged. I also tell it to keep the work inside this project and avoid

03:46 adding dependencies because this example should only need a small change to the calculation. I think this is a useful way to write these prompts. You’re giving the agent the problem, the expected behavior, and a way to check its work. It has enough information to act without having to guess what success looks like. So, let’s send it and see. Okay, it starts by exploring the project files. Then it runs the existing tests and reproduces the failures. You can see it reports that four out of five tests are failing. Next, it edits the function and runs the tests again. And now all five pass.

04:17 I also re-ran the tests separately after it finished and got the same result. Five passed, zero failed. So, that’s great. Now, if we open the change, it’s pretty easy to follow. It converts the price and quantity into numbers, multiplies them together for each item, and adds the result to the total. Then it rounds the final result to cents. The change is two lines in the function. The test file stays the same. That last part matters because we’re checking the fix against the expectations we started with. The string inputs now produce a number. Quantities are included. And the

04:48 empty cart still returns zero. For the multi-item example, two items at $12.50 plus another item at $5.25 now give us $30.25. That matches the expected result. So, this is a useful result from free mode. It handled the file reading, the test run, the edit, and the verification in one request. Now, this is a small prepared example. It shows how free mode handled this particular bug. And I’d want to try more demanding tasks before judging its performance on

05:19 a large project. But for this focus change, it did what I asked. Let’s talk about the allowance next because this is something you should understand before planning your work around it. According to Veridant, free mode refreshes every 5 hours and also has a weekly cap. So, you have both a shorter usage window and a limit across the week. That weekly cap still applies even though the shorter allowance refreshes. Keep both in mind if you’re planning several sessions over a few days. The announcement doesn’t give a fixed number of messages, so I wouldn’t tell you that everyone gets a particular

05:51 number of prompts. Check the usage information in your app as you work, especially before starting something that might need several rounds of changes. For me, that makes focus tasks a sensible starting point. If I already have an error message and a clear expected result, I can put those in the first request. That gives the agent useful context immediately and can save some back and forth. Now, let’s also clear up the difference between free and eco because the zero credit wording can make these sound similar. Free is available without subscribing. Eco also costs zero credits, but requires a

06:24 subscription and has its own quota. Prime spends credits for access to stronger models. So, when you’re trying Verdin without a paid plan, choose free. If you already subscribe, eco is another option to consider for your work. And with prime, the credit cost becomes part of the decision. I would look at the result I’m getting before deciding to move up. If a focus task is being handled well, that’s useful information. If I keep getting incomplete changes or spending several attempts correcting the same misunderstanding, then I would reassess the task, the prompt, and the mode I’m

06:55 using. Anyway, beyond this demo, there are a few other kinds of tasks I would try with free mode. The first is understanding unfamiliar code. Maybe you’ve downloaded a small project and you’re trying to figure out where a form submission goes. I’d ask the agent to trace that flow and explain the relevant files before changing anything. That gives you something specific to check. You can open the files it mentions and follow the explanation. If you’re learning, you can then ask about the particular part you don’t understand. Another task I’d try is a small clearly described interface change. Maybe a button label is

07:29 confusing or an empty state needs a better explanation. I’d give it the exact wording and point it at the relevant part of the project. For that kind of change, I’d open the app afterward and check it on screen. A passing test doesn’t tell you whether text is getting cut off or whether the result looks right at a smaller window size. Documentation is another possibility. Once a small feature is working, I might ask it to explain how to run it, what inputs it expects, and how to run the tests. Then, I’d follow those instructions myself to see whether someone new could actually use them. Those are examples I would explore next.

08:01 The test we actually ran today was the cart calculation, and that’s the result I’m basing this demonstration on. The useful habit is to decide how you’ll check the outcome before sending the request. For our cart example, we had expected totals, existing tests, and a change small enough to inspect. If [snorts] you’re a student working through a small project, that also gives you an opportunity to learn from the fix. After checking that it works, look at why it works. In our example, understanding the difference between adding numbers and joining strings is something you can use elsewhere in JavaScript. And if you

08:32 already code regularly, I’d use the same approach to evaluate how much help it gives you. Try a task you’d otherwise handle yourself, inspect the output, and see whether the time saved is meaningful for your workflow. Overall, I like that you can now try this without making a subscription decision first. The small bug fix worked, the changes were easy to inspect, and we got the expected test results. The allowance will matter if you’re using it frequently, and the scope of your tasks will matter, too. But if you want to experiment with an AI coding agent on a small project, this is a pretty good place to start. I’ll leave Verdens

09:04 website and the announcement in the description so you can check it out and read the current terms yourself. Overall, it’s pretty cool. Anyway, let me know your thoughts in the comments. If you like this video, consider donating through the Super Thanks [music] option or becoming a member by clicking the join button. Also, give this video a thumbs up and subscribe to my channel. I’ll see you in the next one. Until then, bye. >> [music]

09:37 [music]