@jgoldieseo: Nvidia shrank Qwen 3.6 by 3x. But somehow… it kept the big brain. It fixes bugs, reads giant files, and runs on one GPU. It’s fast, free, and runs on your own PC. #OpenSourceAI #NvidiaAI #QwenAI #AITools

Julian The AI Expert

Open In TikTok:

Region: PH

Monday 29 June 2026 06:30:00 GMT

6927

251

Music

Download

No Watermark .mp4 (0.93MB) No Watermark(HD) .mp4 (0.95MB) Watermark .mp4 (1.88MB) Music .mp3

Comments

Enrico Wolf39 :

23.5 gb only for the weights no kv cache (context) so I wonder how you can claim it fits 1 gpu?

2026-07-01 05:48:39

Pelle :

yeah but you need still to "upload" a 35b model to your graphic card

2026-06-30 16:01:59

marcohuber1992 :

i have it with my m3 max 36 gb , is ok but you can Not compare with Claude Opus 4.8 , Fable

2026-07-05 05:35:32

Hugo B :

free?

2026-06-30 13:34:16

Claude Akabane :

I tried it for new type of MOE, but i'm still using Qwen 3 coder 80B

2026-07-01 21:17:42

mangga.kuning :

qwen is amazing...

2026-07-01 11:20:41

To see more videos from user @jgoldieseo, please go to the Tikwm homepage.

@jgoldieseo: Nvidia shrank Qwen 3.6 by 3x. But somehow… it kept the big brain. It fixes bugs, reads giant files, and runs on one GPU. It’s fast, free, and runs on your own PC. #OpenSourceAI #NvidiaAI #QwenAI #AITools

Julian The AI Expert

Open In TikTok:

Region: PH

Monday 29 June 2026 06:30:00 GMT

Music

Download

Comments

Enrico Wolf39 :

23.5 gb only for the weights no kv cache (context) so I wonder how you can claim it fits 1 gpu?

Pelle :

yeah but you need still to "upload" a 35b model to your graphic card

marcohuber1992 :

i have it with my m3 max 36 gb , is ok but you can Not compare with Claude Opus 4.8 , Fable

Hugo B :

free?

Claude Akabane :

I tried it for new type of MOE, but i'm still using Qwen 3 coder 80B

mangga.kuning :

qwen is amazing...

Other Videos

About

Legal