DeepSeek V4.1 Flash + Opencode = FREE AI Coding!

Julian Goldie (0:00)

You can now use DeepSeek V41 Flash for free inside OpenCode. Now, if you're wondering how to get access to that, let me show you today. So you can just use OpenCode, you can get it from their website, or we've already plugged it into our Gentic operating system. And then inside OpenCode here, we can select between the different free models, and you can see that we have DeepSeek V41 Flash for free available right here. Also, there's another new model that is free, which is Nemetron 3.5 Super VL. So you can use this multiple ways. I'll come on to that in a second. And let's say, for example, we wanted to build something out, we can just click a prompt like this, and we're straight into it. The other cool thing about this is, for example, we have OpenCode as a harness, but then if we switch over to the workspace, we can see everything that we've built and preview it inside this section too, which is pretty cool. Now also, OpenCode just dropped in so many cool models. So for example, Cat Jam just dropped, and that's like a crazy fast stealth model that they haven't revealed who's behind it recently. You've got Nemetron 3.5 VL that just dropped today, and then you have DeepSeek V41 Flash, which is like a seriously good model with OpenCode. Now it is, you know, as a limitation, apparently it's only 200,000 Token Context window. For most people, they're not going to use more context than that on the day to day.

So pretty cool that you can get a free model right there. And then also what you can do here is you can go over to Hermes, and inside Hermes, you can use DeepSeek V41 Flash here too for free. So we can just test that's working right now. And then we have DeepSeek working in two different ways, right? And you can see here, four steps done, zero dollars cost, 48 seconds to build this. If we open up, we've got our build over here, so you can see it's all actually working, which is amazing. So super powerful stuff, really easy to set up. And you can see that it works inside Hermes as well.

The fact that you can use it inside two different agenta canisters as a free model is awesome. You may say, oh, once if I get rate limited, blah, blah, blah. See, there's loads of free models you could switch to in between if you need to switch in between these. So for example, you could switch to Nemetron 3.5 Ultra, Long Cat 2.5 Preview. There's like loads of different free models, Big Pea, Cat Jam, etc. And then one thing that you can also do over here is you can use DeepSeek harness. The DeepSeek harness, you can start using like so. And we've actually got two ways to use DeepSeek. So we have the coder here, and then we have the harness. The coder is really cool because you can code on the left-hand side, see your preview here, and then you've got the save builds over here. And then inside the harness, you can actually start using the agentic harness with DeepSeek directly here. And you just have to start it up, boot it up, and then it's good to go. Bear in mind, the agentic harness from DeepSeek is free, and then you can use the free brain, which is DeepSeek v4.1 Flash.

Now, the other cool thing about DeepSeek harness with DeepSeek v4.1 is the brain, and you can get open code to brain and plug that in, is that you can basically build it as you go along. So everything inside DeepSeek is ready to go at one point. And also, what I like about this whole system is you can plug it into a memory here. So let's say, for example, I haven't used DeepSeek for probably about two to three months, and so it doesn't have much information or context for me normally. But what I can do is plug it into my memory galaxy, like you can see right here, and then we can start using the memory galaxy to give it context. So for example, every few minutes, you can see one minute ago, two minutes ago, open code log, it plugs in all of its builds directly into my memory system so that all of my AI agents inside the memory galaxy have an understanding of what I'm doing, what I'm working on, etc. And then that works across DeepSeek Harness, across Hermes Agent, across open code, and we're ready to go every time we use it, which is pretty awesome. Now, the limitation of this, I would say, is a DeepSeek v4.1 flash is not its pro model, so it's not like the same frontier level, but it is much more improved than their original DeepSeek v4. And also, this is not designed to be something that replaces Opus 5.5 or any frontier models like that, but it is great agentically because most of the time, you don't need a frontier model. And number two, you can code and build out with it as well. Now, the other thing I would say here is like for me personally, my genetic countesses are going to be stuff like Hermes and Claude for the main stuff, but I can pass that over to something like OpenCode later. And so the way that I would recommend using this is you have Claude and Hermes as the key orchestrators. But then if you need a subagent that's really fast, or you need to use like a faster option inside Hermes agent, you can switch over to DeepSeek v4.1 Flash anytime as well. Now, if you're wondering how to get access to OpenCode, you can just run this terminal command inside the terminal so you can copy and paste out, or you can just download and install it like so. It's also available on GitHub, and you can see here that it's already got 214,000 stars. So it's a pretty legit agentic harness. The other thing is that it's designed to be token efficient. One thing to know about it though is that, as far as I'm aware, Alibaba actually built it. Obviously, Chinese company, so just something to be aware of there. But for me personally, I like to use it inside my systems. It's also just a good backup option whenever we need it. We can also use it inside other workflows. So for example, you can have a separate profile like we've got for DeepSeek V41 Flash inside HermesBots, and then you can use it in the same way you would use Groggbot. You could plug it into your voice agent. You could have it inside Hermes Oracle, Hermes Muse, Hermes Astrost, and all these other custom workflows, even our lead generation tool right here, so that you can build amazing custom workflows in between open code and the other agents. We even have, for example, a pop-up voice agent on the right-hand side that we can speak to at any time. And this is just all controlled inside our mission control here. So thanks so much for watching. If you want to get more training on this sort of stuff, feel free to check out the AI Profit Boardroom.