I Am #AliveAndThriving Today! || #IAmAliveChallenge || Daily Off-Grid Jungle Journal Entry: Day #1421 - Preparing to Run and Test Qwen3.6 35B A3B on 8GB of VRAM and 64GB RAM - Lower Puna, Far East Big Island, Hawai'i - Tuesday, June 9, 2026
I live naked and barefooted, very close to Earth and Nature, in an 18-acre, off-grid, clothing-optional, food-forest intentional community (GaiaYoga Gardens), way out in the jungles of Lower Puna, far East Big Island, Hawai'i, and I have for more than 7 years now. Although there are many challenges, I love my life, and I'm immensely grateful to live where and how I do, on my own terms! I would not want to live any other way! 😁🙏💚⚡💥🔥✴️✳️❇️👣🌱✨🤙
Hive Debt: 26.47%
HIVE Fiat-Dollar Price: 0.047 USD
HBD/HIVE-Backed Dollar Fiat-Dollar Price: 0.945 USD
ALIVE SWAP.HIVE Price: 0.02192823 SWAP.HIVE
HEARTBEAT ALIVE Price: 0.11028323 ALIVE
BLURT Fiat-Dollar Price: 0.00075435 USD
QORT Fiat-Dollar Price: 0.133378 USD
VOI Fiat-Dollar Price: 0.0001494 USD
BTC/Bitcoin Fiat-Dollar Price: 61,662.60 USD
BCH/Bitcoin Cash Fiat-Dollar Price: 197.08 USD
LTC/Litecoin Fiat-Dollar Price: 41.775 USD
Gold Fiat-Dollar Price: 4,101.88 USD
Silver Fiat-Dollar Price: 64.54 USD
Ranking for Our @aliveandmmb Witness Node: #125 (among all witnesses, including inactive)/#99 (among active witnesses only)/652.8k HP in votes/50 votes in total
Warm greetings all! 😁🙏💚✨🤙
As I mentioned in my post yesterday, through a good bit of online research, I've discovered effeftive methods of successfully running quite large AI models, such as Qwen3.6 35B A3B (35 billion total parameters, 3 billion active parameters for each token) on consumer-grade desktops and laptops with 8GB of VRAM (even on 6GB of VRAM), which until recently was quite impossible. That's how quickly AI and agentic-AI tech is developing and evolving. I now have the information necessary to test this in practical terms.
The model that I've been using up to this point, two variations of Gemma 4 E4B, has 7.5 billion parameters, of which 4 billion are effective for each token, so going from that to a much larger MoE (Mixture of Experts) model will be a very signifant upgrade to the practical capability of my local agentic-AI system. Whether this will end up being more then four and a half times better remains to be seen, but given how capable a model that Qwen3.6 is, this does seem somewhat likely.
Like Gemma 4 E4B, Qwen3.6 35B A3B is a multimodal model with the inbuilt capabilities of vision, reasoning, and tool use, so it should be able to do everything there Gemma 4 E4B can do, along with quite a lot more, and in a notably more effective and efficient manner. As an LLM for my Hermes Agent setup, Qwen 3.6 35B A3B would work much better indeed, if I'm able to get it up and running successfully.
I still need to synthhesize the research information on the important llama.cpp run flags, and to decide on exactly which version of llama.cpp to use, as along with the official version, there are also multiple community forks, with different capabilities, features, and specialized optimizations. Pretty much any of them should work for me, but I want to use the one that allows me to squeeze out the highest performance and stability, while using the least amount of system resources.
I've already downloaded the Q4_K_M quantization of Qwen3.6 35B A3B, which is just over 22GB, though I will also likely test versions with different quantization techniques and formats, as some strongly decrease VRAM usage, while also not losing much model quality. Once I decide on which iteration of llama.cpp, on the optimal quantization type and size, and on exactly which flags and their values to use (I have a few possibilities), I'll be good to go for my first test runs. Hopefully I'll be able to get through all this before the end of the week. We shall see indeed.
Yesterday, Tuesday, I left the Flow House, after finishing up my Hive-Blurt-'We Are Alive Tribe' tasks, by a little after 2:30PM, giving my attention first to some photography for my posts, and to a few small jungle-community tasks, including chopping and dumping bad/old coconuts at the upper coco bar, and potting up a cutting of a rarer species of Plumeria (frangipani for my European friends), for Melekai to plant down near the beautiful space that's she's in the process of creating. With those completed, I took a quick break up at the Flow House, before diving into digging another round or planting holes, this time down next to Kana's place, with a shovel and an o-o bar. When I finally wrapped that up for the day, I took a shower, made myself some food in Wabisabi, collected my leftover superfood fire coffee from the fridge, and with those in hands I headed back to the Flow House, to do a bit more agentic-AI research, and to begin my evening Hive tasks. I got through all of my notifications quite early once again, by a bit before 9PM, and after more AI research, I went to bed by about 11:15PM. I woke up by about 5:45AM, having slept well, going to prepare my superfood fire coffee in Wabisabi, then returning to the Flow House to write this post. I took a quick break to go help to lift a new tarp over the top of the Bedouin, heading back to the Flow House immediately afterward, to finish writing my post. It's now just before 12:45PM, so it's an excellent time to wrap this up, so that I may attend to my few other important Hive-Blurt-'We Are Alive Tribe' tasks, before preparing for a session that's scheduled for 1:30PM. I deeply appreciate y'all so very much! Until tomorrow's edition of these posts! Always forward, onward, upward, and whaleward, ho, together we go! 😁 🙏 💚 ✨ 🤙
2026 Life Goals
1.) Do regular integral practices again, alone or with others.
2.) Prepare regular batches of my medicinal teas collected from the land again.
3.) Make my plant nursery, and the gardens, beautiful again.
4.) Begin learning relevant coding/programming languages to more fully contribute to the sovereignty-driven technologies where I am active, like Arch Linux, Hive, Qortal, DeSo, and Bastyon.
All photos were taken with my Motorola G Play 2024 Android Phone.
Thank you all so much who have helped me get to where I am today, and allowing me to share more of the beauty and magic from my life and my world with you, and for your continuous appreciation and support! I am truly deeply grateful! 😁🙏💚✨🤙
If you'd like to find me on other alternative platforms where I have accounts (I spend most of my time here on Hive), click on this signature image below to go to my LinkTree page.
Come join us in the We Are Alive Tribe on Hive and on Blurt!
Image created by @benthomaswwd
You're invited to provide liquidity in our two ALIVE diesel/liquidity pools, both of which offer daily liquidity rewards:
There will be more ALIVE pools coming soon!
Vote for our @aliveandmmb Witness Node!
This is our collaboratively shared Witness node between the We Are Alive Tribe, and @borniet's and my MMB (Maximum Mutual Benefit) Project, here to build something good for all of us! The higher in the witness ranking we go, the more that we'll be able to grow the We Are Alive Tribe, and the more that we'll be able to share with all of you!
If you appreciate what we're doing with the We Are Alive Tribe, and feel inspired and motivated to lend a hand in building and growing our decentralized community economy and financial framework, which benefits all of us, community conntributions/donations/investments may be sent to our 'We Are Alive Tribe Sovereign Wealth Investment Fund' account, @waat-swif. They're deeply appreciated, and are put to very good use.
Vote for @borniet's @botlord Witness Node!
Bjorn is our badass coder who hosts and manages our (and also Bradley's) community backends, he creates all sorts of useful tools, and he assists people all over the Blockchain constantly. He redefines the term 'busy'!
Vote for @bradleyarrow's 
Witness Node!
Bradley's tenaciously committed and all-in on Hive for the long haul (just like me!), and the higher his witness goes in the ranking, the higher the weekly BBH and BBHO dividend payouts will be!
Join the worldwide Satsman Army! It's by far the best, most fun, inspiring, and motivating way to earn Bitcoin/BTC!
Signature image created by @doze, and the dividers made by @thepeakstudio, with all tweaked to their present form by me.
!BEER
Multitudinous mountains of thanks, my friend! 😁🙏💚✨🤙
!BEER
!ALIVE !LOLZ !BBH
Thanks so much, my friend! 😁🙏💚✨🤙
!ALIVE
!BBH
!LOLZ
Great post as always!
I have a random thought; it's probably nothing.
Qwen can do more for less cost in many cases, but I worry about security and doing anything to give an authoritative state an edge in the AI competition. Any thoughts on my random thought?
Some research I did:
Source: https://www.uscc.gov/sites/default/files/2026-03/Two_Loops--How_Chinas_Open_AI_Strategy_Reinforces_Its_Industrial_Dominance.pdf
I may be overly cautious but I don't want to be doing anything to increase China's dominance in anything, especially AI and robots. But that's just me! And I'm sure I own lots and lots of things Made In China so I'm not that concerned after all, maybe! :D
Happy middle of the week! Hang in there, the weekend is coming. :)
!ALIVE !BBH !UNI !PIZZA !LADY
View or trade
LOHtokens.@kenny-crane, you successfully shared 0.1000 LOH with @tydynrain and you earned 0.1000 LOH as tips. (2/11 calls)
Use !LADY command to share LOH! More details available in this post.
Thank you very much, Kenny, I appreciate it! I hadn't heard about specifically about Chinese models being more susceptible to cyber risks, but I'll look into it a bit more closely. The good thing about open-weight models, which is equivalent to software's open source, is that there's nothing hidden, and they can be modified, such as with various uncensored iterations. Also, China is already way ahead in AI development. Chinese people make up a huge portion of the AI-development community, so it's not even just at the governmental level. One reason that Chinese models are used so extensively is because they are generally better and more capable. Google's Gemma 4 line is very good, and I use that too, but for the most part American-made models can't compare to Chinese models in capabilities and performance, at least not yet. Regarding the exposure of sensitive data, that would be an issue with any cloud-based model that has access to a device, such as through an agent, because anything it sees and processes goes through, and is recorded on the hosting servers. Aside from not being willing to pay to use AI, that is my main reason for wanting to run everything locally, under my own control. 😁🙏💚✨🤙
!ALIVE
!BBH
!LADY
!PIZZA
!UNICOIN
That AI model is used by #BraveBrowser and its "sister" project #BraveSearch. 🤔🧘♂️ The company explained that such model is efficient. 🤓😅
!LOLZ
Are you talking about Gemma 4? I haven't looked into the model or models that they're using yet. 😁🙏💚✨🤙
!LOLZ
No, I was talking about #Qwen !INDEED. 🤔🤯 The #Brave team says that such model is still accurate despite its relatively smaller size than the other popular LLMs. 🤓😏
!WEIRD
!HOPE
!WINEX
Ah, OK, I see. Yes, for its size, Qwen3.6 punches way above its weight class, depending on the specific size and quantization. I can't wait to finally get Qwen running successfully. The larger Gemma 4 is running fine, though loading Qwen still has issues. 😁🙏💚✨🤙
!ALIVE
!BBH
!HOPR
!INDEED
!PIXY
Let's !HOPE that such LLM has less security risks over time !INDEED! 🤗😊 One way for a user to reduce such risks is to make the chatbot completely offline, though it would reduce the effectiveness of the model. 🤔🤯🤓
!WEIRD
!WINEX
Well, I was finally successful in getting Qwen3.6 35B-A3B APEX IQuality to run using both ik_llama.cpp and llama-cpp-turboquant. Now I just need to tweak and refine the flags a bit more. A totally offline model is not feasible for me, as it needs to be able to do web searches and synthesis, and to connect and transact within multiple blockchain ecosystems, Hive included. 😁🙏💚✨🤙
!ALIVE
!BBH
!PIZZA
!PIXY
!UNICOIN
For those who prioritize security, such LLMs should be used on a separate computer (which may be expensive) or in a virtual machine (which will reduce its efficiency) !INDEED. 🤔🤯🤓
!WEIRD
!HOPE
!PIZZA
There are multiple ways to prioritize security without crippling your setup. 😁🙏💚✨🤙
!ALIVE
!BBH
!PIXY
!PIZZA
Yes, and there would be pros and cons of each !INDEED, which an experienced user would barely have any problem choosing from. 🤔🧘♂️😎🤓
!HOPE
!WEIRD
!PIZZA
I'm far more concerned with the idea of important private data being accessed, passing through, and being recorded on the servers where cloud-based frontier models are hosted. That's a big reason, aside from being completely unwilling to pay to use them, that I choose to work with only locally-hosted models. 😁🙏💚✨🤙
!HOPE
!INDEED
!PIZZA
!WEIRD
I was thinking about locally-hosted AI LLMs, despite not being on cloud servers, send ("unintentionally" or otherwise) personal data on the machine to somewhere else on the Internet !INDEED. 🤔🤯 It is already a given on cloud-based AI LLMs that users should care to avoid putting any personal or sensitive information there. 🤯🤓
!WEIRD
!HOPE
!WINEX
For locally-hosted models, there's really no way for them to interact with the outside world unless it's via an agent harness, like OpenClaw, Hermes, Pi, Mercury, or any of the other agentic frameworks, since they give LLMs the tools and capabilities to do so. Because that's the case, strict controls can be placed on how they do so. 😁🙏💚✨🤙
!HOPE
!INDEED
!WINE
!WEIRD
That seems to be like a web proxy which an app (the AI chatbot in this case) is required to use before being able to access the Internet (for the chatbot to perform web searches in this case) !INDEED. 🤓
!HOPE !WEIRD !ALIVE !PIXY
Agent harnesses are used in place of chatbots, and have vastly more capability, and whoever is running them is almost always the one who set them up, either locally, or in a VPS, so there's no 'web proxy' about it. 😁🙏💚✨🤙
!ALIVE
!HOPE
!INDEED
!PIXY
!WEIRD
It really looks like you are talking not only from knowledge but personal experience !INDEED! 😎
I really think that I should experience agentic AI myself in order to fully understand your explanation. 🤯😅 Meanwhile, your comments there stay on the Hive blockchain for me and other people to come back to sooner or later. 😏
!WEIRD
!HOPE
!ALIVE
Indeed, I do research in this area very often, and I also run, and am building my own agentic-AI framework with Hermes Agent, which is the most capable, and the most amazing agentic framework that I've ever seen.
Yes, you should, as it's well worth the time and effort. That's the beauty of publishing posts and comments on a blockchain. 😁🙏💚✨🤙
!ALIVE
!HOPE
!INDEED
!WEIRD
When you have already "finished" your work with your agentic AI (or more accurately, having learned enough of it), you might run your own Hive bots and nodes yourself too! 🤗
!PIXY
Perhaps, though I've noticed that bots run by AI agents don't tend to fare well here on Hive, especially if they make posts or comments. 😁🙏💚✨🤙
!PIXY
Perhaps those bots and nodes that you would eventually host would be fine here on Hive if they do not publish posts or comments !INDEED. 🤖🧘♂️🤓
!PIXY
!ALIVE
!UNICOIN
When I finally get my chosen models up and running consistently well with Hermes over the long term, I'll likely have it assist me in managing assets, and possibly finding particular kinds of posts. 😁🙏💚✨🤙
!ALIVE
!BBH
!INDEED
!PIXY
!UNICOIN
Your agentic AI might then recommend you to play certain games here on Hive as a form of investing your Hive-based tokens! 🎮😅
!ALIVE
!PIXY
Once it gets sufficient information and context, I imagine that it would suggest all sorts of things indeed. 😁🙏💚✨🤙
!ALIVE
!BBH
!PIXY
!UNICOIN
You would then "teach" your agentic AI to make more suggestions that you actually like and less of those that you don't !INDEED. 🤓
!WEIRD
!HOPE
!WINEX
Congratulations, @savvyplayer You Successfully Shared 0.100 WINEX With @tydynrain.
You Earned 0.100 WINEX As Curation Reward.
You Utilized 1/4 Successful Calls.
Contact Us : WINEX Token Discord Channel
WINEX Current Market Price : 0.030
Swap Your Hive <=> Swap.Hive With Industry Lowest Fee or Highest Reward : Click This Link
Read Latest Updates Or Contact Us
Given that Hermes Agent remembers, learns, and improves over time, yes, that is very likely. 😁🙏💚✨🤙
!ALIVE
!BBH
!PIXY
!UNICOIN
I think that you mean that your Hermes agentic AI will know better about your preferences without you (directly) telling about it. 🧘♂️
Anyway, I think that's the case with most locally-hosted AIs with persistent memory. 🤯😅
!PIXY
!PEPE
That's just another way of saying the same thing.
Partly true, but Hermes takes that to a whole new level indeed. 😁🙏💚✨🤙
!BBH
!PEPE
!PIXY
I predict that your success with agentic AI would be noticeable here on Hive (especially on your posts and comments)! 🤖😎
!PIXY
I'll create what I want in time, though I won't use it for creating posts or comments. I'll always do that manually. 😁🙏💚✨🤙
!ALIVE
!BBH
!PIXY
!UNICOIN
I assume that you would sometimes use your agentic AI for understanding posts and comments (especially hard-to-understand ones). 🤓
!LOLZ
There aren't many posts that I don't understand, unless they're in a language that I don''t know. For posts, I'd likely use it to find ones on different topics. 😁🙏💚✨🤙
!LOLZ
Brilliant answer, thanks!
I knew you would have something important to say on this, as you think carefully about everything that you do. Thanks again for the interaction, I appreciate you, my friend!
!ALIVE !BBH !UNI !PIZZA !LADY
I'm grateful that you appreciated it, and of course, you're always very welcome! The appreciation is very mutual, Kenny! 😁🙏💚✨🤙
!ALIVE
!BBH
!LADY
!PIXY
!UNICOIN
Good luck with your system testing my friend wishing you success in all you do
@tydynrain
Made in Canva
!ALIVE
This post has been curated by the Alive And Thriving Team, we curate good content in the We Are Alive Tribe that is on topic for #aliveandthriving, and it's included in our daily curation report on @aliveandthriving, plus @youarealive is following our Curation Trail.
Thanks a lot, @benthomaswwd, I appreciate it, brother, and very likewise. 😁🙏💚✨🤙
!ALIVE
!BBH
!LADY
!PIXY
!UNICOIN
!BBH
Thanks so much as always, Bradley! 😁🙏💚✨🤙
!ALIVE
!BBH