Pinned Tweet
Grab your Own1 box at ownx.co/?via=johnny so you host websites, build local AI inference apps, generate images, and voices using open source AI models! You can also run agents 24x7 - more video guides coming soon. @OwnXOfficial

NEXT VIDEO: Running a @NEARProtocol Validator kdragon and my upcoming hackathon submission(s) 👀
14
10
26
1,147
John Forfar | AI Engineer Dev Rel retweeted
Hardware and uptime are the hidden costs most people skip. Delegating gets you the yield without the ops work, though the math still depends on stake size. Which NEAR-built apps grew revenue fastest this week, ranked here: x.com/HYPERDailyTK/s…
Top Dapps by % Growth of Revenue 7D ⚡️

▫️@WalletV_io | 559%
▫️@SplashOS | 460%
▫️@onchaincc | 215%
▫️@AxiomExchange | 175%
▫️@MoonBotTrade | 118%
▫️@NEARProtocol | 116%
▫️@rainbowdotme | 90%
▫️@InsilicoTrading | 80%
▫️@MetaMask | 77%
▫️@vergex_ai | 76%
0
2
4
448
I ran a @NEARProtocol node at home for nearly two weeks.

Then I checked what a validator seat actually costs.

About 11,000 NEAR right now. Call it $42,000.

To earn roughly $2-4k a year, before power and internet.

And you can earn almost the same rate from your phone, ardware.

Here's what running one actually takes 👇
15
12
112
11,086
Numbers move: the seat price changes every epoch (9,300-11,300 NEAR this week), and at ~5% a seat earns closer to $2k/year than the $2-4k I said.
Live figures: nearblocks.io/validators
Make sure you check latest epoch numbers! Thanks
0
0
2
464
Replying to @johnforfar
One thing from the video: check the drive.

NEAR recommends 2-3TB minimum for a validator.

Mine is 1TB and testnet alone pulled 160GB in the first 10 days, without even syncing a full shard.

Spec 4TB or 8TB at checkout. @ilblackdragon @mert
1
0
2
695
RT @toly: My ram
My datacenter
0
16
0
20
I gave 8 AI employees a task (@OwnXOfficial Next Team) and left them running overnight - then I benchmarked them: Hermes by @NousResearch vs @zeroclawlabs

One box on my desk. No cloud, no API keys, no per-token bill.

Then they wrote the benchmark report about themselves, ve before it goes live on my blog
15
11
19
471
Same eight deliverables, two agent frameworks, graded by a script.

19 minutes vs 68.
8/8 deliverables vs 6/8.
On 4-12x fewer tokens.

The lesson I didn't expect: give anything that touches numbers a code tool. Without Python an 80 row sales report failed every time on both. With Python, both were perfect.
1
0
2
94
Full write up, every number: johnforfar.com/posts/i-put-a-…

The box, if you want one: ownx.co/?via=johnny (JOHNNY10 for 10% off)
0
0
1
47
For those who want 10% use my coupon code JOHNNY10 at checkout !

ownx.co/?via=johnny
0
0
1
22
Happy Halloween! 👻
10
9
16
292
Don't need to vibe code these 150,000 open source apps that already exist! I walk though a bunch of them on this video. Please like & RT.

Comment "OWN AI" below for my 10% off voucher code for the Own1 HARDWARE checkout.ownx.co/?via=johnny @OwnXOfficial @AshtonHetti s
4
4
10
554
Replying to @johnforfar
Watch my previous video about Nexy OwnOS AI Agent: youtube.com/watch?v=RALE5u…
1
0
1
39
Replying to @johnforfar
Next video I'm going balls deep into @NEARProtocol @near_ai @IronClawAI @near_intents and running a #NEAR validator on the Own1. ALSO don't miss out on the NEAR HACKATHON >>> $250,000 USD prizes to be won starting in a couple of days: nearhackathon.com
1
0
2
59
John Forfar | AI Engineer Dev Rel retweeted
Replying to @Kimi_Moonshot
K3.1 Pro works directly in the tools used by the 3D modeling and game industries - Blender, Unreal Engine, Godot, and CAD.
On BenchCAD, which tests 3D reconstruction from multi-view renderings through CAD code generation, it scores 92.7% with tools, approaching Claude Opus 5.5 (95.9%).
K3.1 Pro can deliver a fairly complete project, from asset construction to scene assembly, animation, and rendering.
17
73
1,978
159,468