Coding with Lewis
Region: CA
Wednesday 14 October 2026 16:35:32 GMT
12697
601
18
43
Music
Download
Comments
Wakarimashita :
How has 32 GB of VRAM at home ?
2026-10-24 02:49:05
0
kevjonz :
I have 32gb ddr5...😭
2026-10-14 19:38:22
0
Johannes Wilm :
actually you can go further than that. With my 91GB RAM and an aorus rtx 5O60 Ti egpu, I can run Qwen Flash Next with a 262 token context window at around 55 t/s. Lookup "Strata".
2026-10-14 17:21:27
4
Glyn :
You skipped the 6 and 4GB vRAM.
2026-10-24 10:31:59
0
Turan :
Genuin question: what is the topic and what is the guy talking about? Sounds interesting
2026-10-14 18:06:34
3
Coding with Lewis :
What one do you rock?
2026-10-14 16:35:55
0
Michelle Traska :
I wish this was in English. 😂
2026-10-16 22:18:13
0
nopenope :
2 3090,s Nvlinked is 48gb vram for around $2300. the omni merge qwen3.8 has the overthinking and looping trained out of the model. its my primary q8 all the way
2026-10-14 22:21:00
0
🤡 :
could you compare it to smth measurable for developers? like the models we use everyday... like is it opus 5.5 lvl or sonnet?
2026-10-14 22:22:22
0
TheZainEditor :
8 gb
2026-10-18 13:47:57
0
Hardt 🇿🇦🇳🇱 :
I have an rtx3060 ti 8gb running qwen 3.8 flash next 125B on strata. I get around 40t/s with 256k context
2026-10-15 08:05:21
2
MAVE-MECHANICS :
not enough parameters. once i can run 600b param locally then talk to me
2026-10-14 19:49:09
0
Yaseen_magdi :
[Big LOL][Big LOL][Big LOL]
2026-10-14 20:38:19
0
To see more videos from user @lewismenelaws, please go to the Tikwm
homepage.
