Would love to see a Qwen 3.5 release in the range of 80-110B which would be perfect for 128GB devices. While Qwen3-Next is 80b, it unfortunately doesn't have a vision encoder.

▲

Tepix an hour ago | parent | next [-]

Have you thought about getting a second 128GB device? Open weights models are rapidly increasing in size, unfortunately.

▲

PlatoIsADisease 2 hours ago | parent | prev [-]

Why 128GB?

At 80B, you could do 2 A6000s.

What device is 128gb?

▲

the_pwner224 2 hours ago | parent | next [-]

AMD Strix Halo / Ryzen AI Max+ (in the Asus Flow Z13 13 inch "gaming" tablet as well as the Framework Desktop) has 128 GB of shared APU memory.

	▲	scoopdewoop 3 minutes ago \| parent \| next [-]
		Not quite. They have 128GB of ram that can be allocated in the BIOS, up to 96GB to the GPU.
	▲	hedgehog 19 minutes ago \| parent \| prev [-]
		Keep in mind most of the Strix Halo machines are limited to 10Gbe networking at best.

▲

lm28469 an hour ago | parent | prev | next [-]

That's the maximum you can get for $3k-$4k with ryzen max+ 395 and apple studio Ms. They're cheaper than dedicated GPUs by far.

▲

tarruda an hour ago | parent | prev | next [-]

Mac Studios or Strix Halo. GPT-OSS 120b, Qwen3-Next, Step 3.5-Flash all work great on a M1 Ultra.

▲

vladovskiy 2 hours ago | parent | prev [-]

Guess, it is mac m series