/tech/ - Soyence and Technology » Thread #41808

[21 / 4 / 1]

84KiB, 403x649, nate-crying-gaping.gif

Chud Sun 11 Oct 2026 09:06:38 No.41808 View View Original Report

>just let the LLMs code dood let LLMs do everything I'm totally not a worthless goy who is falling for a kike psyop

View Same Similar Google ImgOps iqdb SauceNAO 1789480952190w.webp, 1.7KiB, 221x255

Chud Sun 11 Oct 2026 09:09:52 No.41809 Report

just let the LLMs code dood let LLMs do everything I'm totally not a worthless goy who is falling for a kike psyop

View Same Similar Google ImgOps iqdb SauceNAO rumir.jpg, 33KiB, 339x500

Chud Sun 11 Oct 2026 09:35:25 No.41810 Report

>look at me, I'm OP. I spend hours coding uoptimized, trivial snca, which claude could do in 2 minutes and I love to suck COCKS. Find me on bluesky!
don't melt off the stick too fast, we want to enjoy your tantrum

View Same Similar Google ImgOps iqdb SauceNAO smugjak-ear.png, 31KiB, 628x772

Chud Mon 12 Oct 2026 14:00:04 No.41907 Report

>>look at me, I'm OP. I spend hours coding uoptimized, trivial snca, which claude could do in 2 minutes and I love to suck COCKS. Find me on bluesky!
>don't melt off the stick too fast, we want to enjoy your tantrum

Chud Mon 12 Oct 2026 14:03:35 No.41909 Report

Quoted By: >>41911 >>41915 >>41917 >>41946

Use chink local ai models to set yourself free from jews and still take advantage of the tech

Chud Mon 12 Oct 2026 14:12:40 No.41911 Report

Quoted By: >>41922 >>41926

>>41909
I actually did this and just made an inbred version of Grok

Chud Mon 12 Oct 2026 14:20:54 No.41915 Report

>>41909
how do I make this work on pentium 2 pc

Chud Mon 12 Oct 2026 14:24:30 No.41917 Report

Quoted By: >>41926

>>41909
>cede all your coding ability and freedom to a machine because you're lazy and incompetent
Yeah I'm not doing that.

Chud Mon 12 Oct 2026 14:28:50 No.41922 Report

>>41911
Appalachian grok award

Chud Mon 12 Oct 2026 14:32:43 No.41926 Report

>>41911
das giga
>>41917
Don't do that obviously, only use ai for time wasting tasks nophono even cares about

Chud Mon 12 Oct 2026 19:30:50 No.41946 Report

Quoted By: >>41947

>>41909
What local models/computing power is considered good enough to not be unreliable? I have a 8gb vram gpu and 32gb ddrr4 ram but im scared any model i run will just spew out bullshit that will require hours of debugging

Chud Mon 12 Oct 2026 19:37:09 No.41947 Report

Quoted By: >>41948 >>42040 >>42049 >>42057

>>41946
all local models are garbage compared to cloud frontier models, even if you buy a high-end card like a 5090 for $5 gorillion dollars it won't even compare, and with that money you would be better off buying like 2 years of codex/claude $200 plan, which will get better as the tech improves

Chud Mon 12 Oct 2026 19:44:37 No.41948 Report

Quoted By: >>42058

>>41947
I was planning on getting a claude subscription to help me trough uni and saving up for a powerful machine in the future if local tech improves anyways, thanks for the advice doe

Chud Tue 13 Oct 2026 13:23:52 No.42040 Report

>>41947
>soyim actually spend their monies on kikeware

Chud Tue 13 Oct 2026 15:47:59 No.42049 Report

>>41947
>sunk cost fallacy nigger getting scammed for $200 a month and produce fuck all
>could just buy several full games at full price instead

Chud Tue 13 Oct 2026 16:03:33 No.42057 Report

Quoted By: >>42064

>>41947
small models will get exponentially better
For like 2k you could buy some amd instincts and get about 196gb of vram
Pair that with like 512gb of ram and you are super well off

Chud Tue 13 Oct 2026 16:04:44 No.42058 Report

Quoted By: >>42059

>>41948
LOCAL tech is improving insanely now, you only need enough vram for the active experts now kek
You can run glm 5.3 flash on a system with like 16gb of vram and 160gb+ of ram kek

Chud Tue 13 Oct 2026 16:05:00 No.42059 Report

>>42058
With good decode and prefill speeds too
(like 60t/s dec)

Chud Tue 13 Oct 2026 16:41:40 No.42064 Report

Quoted By: >>42065

>>42057
bus link IO bottleneck doe

Chud Tue 13 Oct 2026 16:42:24 No.42065 Report

>>42064
Thats actually not a bottleneck at all really
You get almost native speeds

Chud Tue 13 Oct 2026 16:45:02 No.42066 Report

Its essentially streaming the experts to vram (ahead of time obviously)
With 8gb of vram and 48gb of ram, I can run qwen3.8-flash-next-gsq-rco-coder (iq3_s gsq-rco (eqv to q4km, fucking crazy kek) with the non code experts stripped, maintains most of the performance)
If i had 16gb more ram, 100% could run the full iq3_S
Qwen3.8-flash-next eqv grok 4.6 according to all sorts of benchmarks and opus 4.7 / 4.8 (depends)

Capcode	All Only Moderator Posts Only Manager Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Embed Type	All Only With YouTube Only With Vocaroos Only With Sounds Only With Flashes
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Post Type	All Only Sticky Threads Only Locked Threads Only Looped Threads Only Opening Posts Only Reply Posts
Data Source	All From archived.st From Log Warehouse From Internet Archive
Results	All Grouped By Threads
Favorites	All Only My Favorites
Order	Latest Posts First Oldest Posts First AI Most Relevant First

On these boards

Your latest searches