All Episodes
Previous Next
Oct. 14, 2025 - Health Ranger - Mike Adams
30:12
BREAKING: Brighteon's new standalone AI engine can be downloaded for free NOW
| Copy link to current segment

Time Text
Okay, welcome.
This is Mike Adams with the biggest announcement of the year by far from us anyway.
And it's an extraordinary announcement, major game changer, that as you'll see, it it changes everything and it will vastly enhance your life.
And it's free.
And we kept our promise, just like we said, and I'll go ahead and explain what this is.
So two years ago, I began diving into building AI engines for human freedom and preserving human knowledge, kind of the Noah's arc of hidden knowledge.
And we released, I don't know, a year and a half ago, we released our first engine, which was called NEO, and yeah, it was okay, it kind of sucked.
And we just didn't have the training down for it.
We didn't have a good data set at that time, etc.
So fast forward then to about a year ago, I promised that we would release a brand new standalone engine that you could download for free and you could use locally.
I promised then that we would release that in March of 2025.
Well, that deadline came and went.
And here it is seven months later.
The good news is it's here now.
We've got it.
And it's already been posted, and you can download it right now.
I'll tell you how here in a second.
So let me just back up and explain uh why it took seven additional months to get to this point.
So, yes, we had models in March and and they sucked.
And we we had models in May and they sucked, uh, etc.
And you know, it was bizarre.
We were not getting the training that we wanted.
And I was talking about this over the summer.
I mean, we weren't getting the training results that we wanted uh initially.
We were getting about 50% alignment in the standalone models, which again, you can download these for free right now.
I'll tell you where.
I mean, let me just say, just go to bright you.ai and click on the downloads link.
It's at the top of the page.
And right there, you can download these models for free.
Uh if it asks you to log in, just enter your email address.
It can be a new email.
It can be, you know, you can just register with a new email.
It'll email you a pin code, you type that in, boom, you're in, and then you can download the models for free.
And uh right now there's only one model linked, but we're gonna link a second one here shortly.
And the only difference is the quantization levels of those models.
But this model has over 95% alignment.
So this is the model that I I had hoped to have uh seven months ago.
And let me back up and explain what it took to get here, because it's it's over two million dollars of expenditure.
And on top of that, it took me uh literally these two years to do the data pipeline processing for all the training and uh the data curation in order to train all these models that we have now to train them on uh natural health information, you know, natural cures, uh, to get rid of the big pharma bias that's in these models and to program them correctly.
Basically, we had to mind wipe the underlying models and retrain them.
And now uh we have been able to do so with extraordinary success.
Again, over 95% alignment.
And it's just extraordinary.
I I've already given this model to uh three people that uh one of them is um a top-level engineer, and uh uh one is uh like a computer science major, etc.
They're just blown away.
And the extraordinary thing about this model is that you can you can download it, use it for free, unlimited.
It's offered under the very permissive MIT license, which means you can use it commercially or non-commercially, and you don't owe us anything to, you know, you don't have to pay for it.
We kept our promise, like I said.
Uh we funded this because of your support for our online store, HealthRangerstore.com, and the Health Ranger Store made a grant to our nonprofit consumer wellness center, the CWC, and then the CWC built this model and released it from the nonprofit.
And so that's why this model is completely free and completely non-commercial.
But uh it's gonna change your life because you can ask it anything.
Anything about emergency first aid, about home gardening, about survival, about preparedness, how to make herbal extracts.
You can ask it things about money, gold and silver, dollars, the Federal Reserve, history, false flag events, vaccines, mRNA, you name it.
You can ask it any of those questions, and you're just going to be stunned at the quality of the answers, and also the different tasks that you can perform with this model.
You can ask it to expand one sentence into an entire article.
It will write articles for you in seconds.
You can ask it to summarize a science paper.
You can ask it to translate an entire article from English into Spanish.
In fact, it supports a multitude of languages.
And we want to give credit also to the Meastral Company out of France, because after lots and lots of testing, we found that the Mistral Open Nemo model, which is a 12 billion parameter model, was by far the best model for this purpose.
And for whatever reason, it turns out that the Mistral Nemo model is very trainable.
Well, actually, of course, we came up with some really clever training methods, signal-to-noise ratio selection, you know, way beyond QLORA as it's called, um mind wipe techniques, but those techniques worked better on Mistral Nemo than on any other model.
And yeah, we tried lots of different models and everything was failing.
At least we couldn't get above 50% alignment.
But with Mistral, we got to 95% alignment, which is just as good as our online version.
So we couldn't do this without the Mistral Company.
And again, I want to publicly thank the Meastral Company for offering these language models open source so that other companies can train on top of them like we are doing.
And as a result, then we're able to get you this free of charge.
We're kind of paying it forward, you know.
Uh we're we're taking this model that Meestral offered for free, and then we're enhancing it with our knowledge, uh, extensive knowledge, I should add, you know, hundreds of millions of pages of knowledge content, and then we're giving that to you for free.
And the reason we're doing this is because, as you may know, uh my mission, this is my quote, moonshot mission, is to change the world by reaching one billion people with a message of empowerment and bypassing censorship, teaching people about natural medicine to help end human suffering, and to help end uh you know diseases, to help end censorship and tyranny and authoritarianism, etc.
And the best way I can do that is to build something like this and give it to the world for free.
So all you gotta do is go to bright you.ai, click on the downloads link at the top of the page, download the model.
It's a G G UF file.
And I have no idea what G G UF stands for.
Someone knows what that means, but it that has eluded me this entire time.
Anyway, you download the file.
It's about a seven gig download for the smaller version, which is the four-bit quant.
You download that file.
Okay.
That file is not executable.
You gotta load it into a piece of software that you can download for free called LM Studio.
Or you can use other software like OLAM, and there's other inference software that's available.
But the best one I found is called LM Studio, which means language model studio.
So you just go to LM Studio dot AI, and you download the software.
It's free.
You install the software, and then you have to take that GGUF file and you have to place it in the proper folder where you store your local uh models.
And that's easy to find out because you can just click on the model icon on the left side, and it shows you where your models are stored on your local drive.
You just go there, create a new folder, uh, call it like CWC, and then underneath that create another folder with the name of the model, which is the uh uh CWC Meastral Nemo 12B.
You can call it, actually, you can call it whatever you want, and then you put the GGUF file in that folder, and then from there it shows up in your LM Studio software.
And then you can click on it and load it up, and you can set the settings for it.
And I recommend that you use a contact length of 8192, 8,192, which is the sweet spot for this model.
It really rocks at 8192.
And you can set the evaluation batch size up to 2048 if you want.
That seems to work pretty well, or you could even try higher.
These are just some settings inside LM Studio.
You can just ignore them if you want and just run it on default.
But the longer contacts length will actually um do really good things for you.
8192 for contact length.
And you can also, if you know what flash attention is, you can turn that on because that makes it faster.
A little bit like 20 or 30% faster.
And then you load it up and then you can just start chatting with it locally.
And uh using LM Studio that has a local API.
And so if you are a Python coder, you can write code to start querying this language model locally, using your own software and your own GPU.
Uh, yeah, you have to have a GPU in order to um, you know, to run this graphics card, typically like a at least eight gigs of RAM on it, which is very inexpensive these days.
Nvidia has a bunch of these cards.
And from there you're you're off to the races.
Now, you can then unplug your Ethernet cable, disconnect from the internet, and you can still run it all you want.
You can ask it questions, it'll generate articles, everything offline.
And that's been my goal from day one.
I wanted to give you tools that were decentralized, tools that did not require the cloud to work in order for you to use them.
Because right now, if you go to bright you.ai and you use our language model online, well, that's fine.
That's great, you know, until the internet doesn't work, or our model crashes, or the power grid goes offline because of World War III, and you know, there you are.
Like, how do I figure out first aid using garden herbs?
You're like, dang it, I can't get to the internet.
Well, that's when you want to have our model installed locally, and you can run it locally, and it'll even run on laptops and things without GPUs.
It'll just run in the CPU, which is much slower, but it's still doable.
So the thing is just download the model.
It's it's a very valuable file.
Again, it took us two million dollars to create that file, by the way.
So download it's a two million dollar language model, free of charge for you, and just start giving it to people.
You can just put it on thumb drives and hand it out.
You know what you really should do is download LM Studio and put that on the thumb drive with our language model, and just start handing it to people.
It's like use this.
This is amazing.
It it knows everything.
It can answer all your questions about cancer cures and the dangers of mRNA and how to detox spike protein, you know, all that stuff, plus the history of the Federal Reserve and why gold is real money, etc.
What happened in the 2020 election, you know, all these kinds of things.
It's really powerful.
And it's not piracy to download this and copy it and give it to people.
I I want you to do that.
This is the only way we're gonna reach a billion people.
We're gonna have to do it together, right?
So please do that.
Uh download it, copy it, install it, show it to people.
Have like an AI party at your house, you know, show people this thing on your computer, and then when they say, oh my god, that's amazing, they see all the answers like how much did that cost?
You're like, oh no, no, here it is, it's on a thumb drive.
You can have it.
So again, big picture here.
This this is an absolute game changer.
I mean, this is gonna change the world, especially the world of independent media, but also it brings knowledge and information to your fingertips completely free of charge, where no one can center it, no one can control it,
no one can interfere with the knowledge and interject their own narratives About, oh, you gotta take flu shots, you know, or whatever narratives are pushed by the deep state or the vaccine industry or the government, whatever government you're living under, whether you're living under the government of you know Turkey or or China or America or whatever.
No government can interfere with this.
This is the most important archive of human knowledge in the AI form that has ever been created in the history of our world, and it's available to you now for downloading.
So help me share it.
Get it out to as many people as possible.
And once you use this again, you're gonna think it's magic.
You're gonna be like, no way.
That's not possible.
Like if you type in a prompt, here's a prompt you can say, just expand this into an article, colon.
Um cancer cures from fresh produce.
Okay, let's say you type that.
You hit enter.
Sit back and be amazed, as it just blurts out a whole freaking article about all the cancer cures, everything you can imagine in great detail from that one sentence.
And if you want, you can you can be more specific.
You can say, I want you to write as a journalist in full paragraph formats, you know, whatever.
You say, I want you to write a speech, or I want you to write a social media post, make it more friendly and hip sounding, or whatever you want to do.
And it will generate it that way.
And then you can take that whole thing, copy, paste it back in and say, hey, translate this into Spanish.
Boom.
And suddenly it's all in Espanol.
Oh, translate it into Chinese, translate it into French.
You know, uh, it does lots of languages, well, whatever languages that the meestrel base model handles.
We didn't train it languages.
That's that's all to the credit of Mistrel.
But we trained it on, you know, a massive amount of English-based content.
Although in the future upcoming versions, we will be training in different languages natively.
So it'll be an awesome engine in Espanyol, and then also in Chinese, French, German, Italian, Japanese, Korean, etc.
Eventually we'll have it speak Australian.
And but that's it's very that's it, it's a hard language to master for some reason.
I don't know.
AI even has trouble with it.
I think they've been talking to too many people in the outback.
Um I'm kidding, we love all you Australians.
You'll love the engine, but it does use uh American English.
Also, for those of you in the United Kingdom listening, we love you too.
But we use uh American spelling.
And we also pronounce aluminum correctly, in case you're wondering.
Um I can't I can't help but tease everybody, including us.
Okay, bottom line is uh it's here.
It's here.
Finally, it's seven months late, but I didn't want to release a crappy model.
We finally got what is awesome.
We're releasing it now, free of charge, just like I promised.
We kept our word.
We did what we said we were going to do.
This is a milestone for not just us and you and all of you who have supported us.
This is a milestone for humanity.
This is a milestone for the independent media.
And this means from this day forward that the powers that be can never crush this human knowledge.
They can't censor it because there's no central control.
And that's why we release this for free on purpose.
We are all about decentralization of human knowledge.
And this model, 12 billion parameters, this is a significant model.
It's much larger than the 7 billion that we released initially, which was the Neo 7B.
This is 12 billion.
And there's more.
You know, Mistral, for example, has a so-called small model that's 24 billion parameters.
They have a medium model that I think must be, I don't know, 72 billion.
They've got a large model.
I don't even know how many parameters that is.
But as compute becomes more and more capable, you're gonna get larger and larger models.
But even this 12 billion parameter model seems to know everything.
Like everything that you can ask it about, literally anything in the world, but especially the things that we've trained it on, you know, really intense training on nutrition and phytochemistry and food and agriculture and honest money and freedom and liberty and history and uh survival and preparedness techniques.
It can teach you how to disassemble and clean a Glock, by the way.
Yeah.
Except for those of you in Australia and the UK who are not allowed to own them for some reason.
Um there I go, teasing you again, sorry.
Um, but if you ever do own a Glock, you can disassemble it following the instructions in the model.
All right, all right.
Uh in Texas, we'll ask it like, how do I disassemble my 50 caliber semi-auto, you know, super gun?
Because that's what everybody has in Texas, apparently.
Those are called Barrett's, by the way, in case you're wondering.
Okay.
Uh lastly, I do want to say as a disclaimer that this model is not intended to provide you with medical or financial advice.
It's not intended to replace your naturopathic physician or your financial advisor, et cetera.
And you should understand that all AI can make mistakes.
All AIs hallucinate.
So if it spits out certain dates or like book titles or legal case numbers, et cetera, yeah, you shouldn't trust those things.
Because AI is kind of famous for fudging those little specifics.
AI is great at giving you the right big picture, but it'll fudge a case number.
Or it'll get a date wrong, or it'll get somebody's name wrong, or it'll make up quotes that people never said, things like that.
You know, all kinds of fun, entertaining things.
So verify all critical information.
Like if you're asking AI, like, what's a great tincture that I should snort?
And if it comes back and says, Oh, you should mix up oregano and cayenne pepper and snort that.
You know, you should engage your human brain and say, that sounds very painful, actually.
That sounds like a horrible idea.
Maybe, maybe I won't do that.
Or um, or I'll pay somebody else to try it first.
You know, no, um, be mindful of reality when it comes to uh the AI engine.
And don't uh don't make any investments based on AI, because you might ask AI, like, what's the price of gold going to be next year?
And it'll come up with a number, yeah.
It'll probably tell you a number that's lower than what gold is right now, because gold's been going up so rapidly.
So it'll say, man, gold, it might hit 3900, you know.
And right now it's over 4100, because these AI engines are kind of living in the past.
Uh I should say our training information for these models, let's see, it's up to I don't know if this batch included September of this year, but definitely August of this year, in terms of our own articles and podcasts and interviews and things like that.
Our online engine at bright you.ai, that definitely covers up to the end of September.
So it's only about two weeks behind right now.
However, uh AI will very often live in the past.
And you know, you should consider that.
Now, if you're asking it about herbal recipes, you know, it doesn't matter because herbal recipes are timeless.
But if you're asking it to comment on current news and things like that, it might tell you, well, the former president Trump, you know, is gonna do this.
Well, yeah, he was a former president, but he's also the current president, and sometimes AI doesn't realize that.
So you need to be mindful of what AI is good for and what it's not good for.
And if AI produces something, some kind of document for you, if you're gonna use it for anything important, like you know, publishing a story or sending a letter to a law firm or you know, writing your resume or whatever, like you should double check that stuff.
Because otherwise it'll it'll be a clown show of hallucinations, you know.
Trust me, the one you don't check will be the one that gets you in trouble.
Now, one more thing, this is also gonna be a game changer for the independent media.
So one of the reasons I wanted to share this tool with everybody, Is because uh independent media creators uh typically have very low budgets, you know, not paid well, don't have a lot of sponsorships, etc.
And so the mainstream media gets all this money while lying to you.
Uh I wanted to help level the playing field, so this tool is free.
And if you're an independent media person, a writer, a content creator, then you're gonna love the fact that this tool is free and it amplifies your time tremendously.
So you can have it generate a new article every 60 seconds.
Not that every article's gonna be perfect or anything, but it it cuts out the time of the initial article composition if you know how to do prompting correctly.
So when you're using the LM Studio software, you want to adjust both the system prompt and the user prompt.
And by doing those two things, you can make it generate just about any kind of writing style or knowledge uh slant that you want, or you know, if you have a site that's focused on a certain thing, you can really direct the answers in that direction by altering the system prompt and the user prompt.
And if you don't know what those are, uh it's time to learn that.
Very easy to learn.
You can either just find videos, watch videos, or you can you can ask any AI engine to teach you about what are system prompts and what are user prompts, or you can I mean, heck, you can just go to bright you.ai, our free engine, and just start asking at those questions right there.
How do I how do I use LM Studio?
How do I use a system prompt?
How should I write a system prompt?
Uh what you know, what's the best way for me to generate articles that are in my same style and tone, etc.
And you'll find out you know about one-shot prompting and multi-shot prompting and extending the context window and things like that that can be very useful to you.
So, bottom line is from this point forward, from this day forward, you now have the tool, you have access to it free of charge, that encompasses all of human knowledge, it can be run locally without the cloud, requires no permission, requires no cost, requires no logins, just requires local hardware.
You do need a computer with a graphics card, and you need to download this file.
Uh, but other than that, there are no restrictions.
Complete freedom of speech.
Now, there are some basic guardrails on the model, like not to answer questions about how to build bioweapons, stupid things like that, obviously.
But uh, you know, aside from that, it's very much pro freedom of speech, and it it's a game changer.
So every kind of content that you create from here forward, whether it's articles, social media posts, letters, papers, executive summaries, business proposals, you name it, uh, you can use this tool to generate all that.
And it's very, very good.
It also writes some code, by the way.
Uh it's very competent at writing code in like uh Python or batch files or uh JavaScript or whatever.
But I wouldn't use it for that because we didn't train it on any code writing at all.
And whatever is in there is just the latent capabilities from the mistral company.
Uh if you want a code writing engine, you should be using anthropic or what's called clawed code.
And I'm using clawed code for a lot of my data pipeline uh processing scripts, and it's just outstanding.
And that's what's enabled me to get so much more done these days.
All right.
Uh finally, a couple other things.
We've got, of course, our free AI engine at bright you.ai.
Use it now, use it every day.
You'll love it.
Or just use the local version I've been talking about here.
In addition, we have uh a new rendition of our site called censored.news.
And if you go to censored.news, every 30 minutes there's a new audio summary that's generated that reads for you all the news and top trends of the last 24 hours.
And it also spiders 77 websites that are censored.
And you can see the list on on the site, just scroll down, you'll see all the websites that we spider.
And then using AI logic, it generates trends in seven different categories like health, finance, tech, etc.
And it identifies the top trends based on all the stories from the 77 websites.
And then, and you can see all those trends, and you can click on the trends if you want to read a summary of each trend with all the sources and everything.
Or you can click play on the audio player, and it will just play this wonderful warm voice, just reading you the news with a special summary of the news.
And it's between five and ten minutes in length.
And this is the first thing that you might want to do each morning.
I don't know.
I mean, if you want to know what's happening, just turn on censored.news and just click play and then just go about your morning, you know, making your coffee or your smoothie or your pancakes or whatever you're doing, and just listen to the news in the background.
And like everything else that we offer, this is free of charge.
It's completely free.
Yeah, so thank you for your support because we couldn't do it without you.
When you shop with us at HealthRangerstore.com, you help us fund these kinds of tools and projects.
And this is the way that we give back to you.
As I've always promised, I've always said, you know, we we take whatever resources that we are blessed with through your support and God's grace, and we put those resources back into these platforms and technologies to bring you amazing decentralized solutions that empower you and bypass censorship and uplift your life with knowledge.
And we're doing that in spades right now, just like just like I promised we would.
So there's much more to come.
We have game-changing tools that are going to be announced in the days ahead.
We have more language models that are going to be announced now that we've finally nailed down the most effective way to alter these models, which only took a year and a half to figure that out.
Um I know I know like a hundred ways to produce badly trained models.
I'm I've got a lot of experience at producing badly trained models in case you ever want to know.
This is kind of like a Thomas Edison problem.
Like I know hundreds of ways to not make a light bulb, but I know one way to make a light bulb.
I know one way to train models that kick ass.
And we've got that way now.
We've got it nailed down as you can see.
So again, the action item, go to bright you.ai, click on the downloads link at the top, go to the downloads page.
You may have to enter your email address if you are not logged in.
And then once you enter your email address, you'll be able to uh download the models.
So yeah, enjoy.
Thank you for listening.
Mike Adams here, the Health Ranger, and be sure to follow all my work at Brighton.com, naturalnews.com, censored.news, and bright you.ai.
All right, thanks for listening.
Take care.
Power up with our organic whey protein powder, a complete protein packed with amino acids, non-GMO and lab tested for purity.
▲
Export Selection