I love the ASUS ROG Flow Z13 Strix Halo Laptop

Nick AntonaccioAdmin
Sep 07, 2026 at 15:30

#11

With the new price increases for DGX Spark machines (all the GB10 variants), the ASUS ROG Flow Z13 laptops are looking even more attractive. New models with 128Gb shareable RAM and 1Tb HD are currently $3,299.99 on Amazon. In comparison, the Asus GX10s are now nearly twice as much.

DGX Sparks have become my default self-hosted production hardware, because you can't get away with fewer than 2 clustered machines for a model the size of GLM 5.3 Flash, and I really appreciate the unbeatable prefill speeds that Nvidia hardware still dominates in.

But there are plenty of very useful models which run very well on Strix Halo. Qwen 3.8 27b has become the standard, and I still love the MOE Qwen 3.6 35a3 - it's blazing fast on the Z13. Others like Mimo 2.5 and the Gemma 4 models are still practically useful and also very performant for production use. Laguna runs really fast on Strix Halo too, and with Nvidia's massive investment in Laguna, I expect we will see tremendous improvements from that company.

So weigh what you actually need to accomplish with a self-hosted model. For personal use, these still relatively affordable machines are great, and if you need a portable laptop, nothing else comes close for the price.

Nick AntonaccioAdmin
Sep 08, 2026 at 12:16

#12

Qwen 3.8 Flash Next runs fast in IQ3_XXS quant on a single ASUS ROG Flow Z13 Strix Halo machine (I'm using the Unsloth quantization).

The only issue is that this quant appears to severely degrade quality. That's quite a shame because this model is seriously powerful without compression. In fact, it competes with GLM 5.3 Flash in many ways that matter, and the GLM model requires at least something in the order of 2 clustered DGX Sparks worth of hardware, to run at the same quant level.

The other real issue with Qwen 3.8 Flash is that it thinks way too much out of the box - a lot like 27b, but even more ridiculous. Luckily, that can be adjusted.

Remember, this model is just a preview of the Qwen 4 series architecture. If Qwen and the community can get a model of this capability to run in just a wee bit less VRAM, I'd expect that to be the version which makes Strix Halo hardware (and other 128Gb machines like Macs and single DGX Sparks) a real power house, with capabilities close to the current frontier.

Nick AntonaccioAdmin
Sep 14, 2026 at 14:17 (edited, 1 revision)

#13

There are currently 4 like-new ASUS ROG Flow Z13 laptops available on Amazon Resale for $2,849.99:

https://www.amazon.com/gp/product/B0DW238TXK?smid=A2L77EE7U53NWQ&th=1

That's a many months ago price - and about half what a DGX Spark currently costs.

Update: price has gone back up to $3,135, as of 9/14/2026 (even for the resale units)

History

Nick AntonaccioAdmin
Sep 13, 2026 at 12:07 (edited, 1 revision)

#14

Donato Capitella put together everything needed to run Qwen3.8-Flash-Next on AMD Strix Halo, with the best performance yet. He's using the Halogen engine to get 752 tokens/s prefill and 41 tokens/s generation at 32K context, and that configuration completed all 19 tasks on his benchmark tests (18 on the first attempt). That's the best combination of speed and performance yet for any model on his leaderboard, making that configuration of Qwen3.8-Flash-Next the current leading model all-around for Strix Halo:

https://www.youtube.com/watch?v=Nm_zN6RQ_eE

That's pretty freakin amazing results: a seriously capable model (beating Deepseek v4 Flash in Donato's benchmarks), running on a single very portable laptop machine (no clustered hardware), which you can currently purchase for $2850. That's hard to beat right now.

His new toolbox is available at:

https://github.com/kyuz0/ai-toolbox-cockpit

Please keep in mind that these successful results were all achieved with the preview model which shipped with Qwen3.8-Flash-Next. Qwen 4 models running on that same next architecture are expected to be released before the end of the year. That should be something genuinely special for the self-hosting crowd.

History

Nick AntonaccioAdmin
Sep 21, 2026 at 12:42 (edited, 3 revisions)

#15

There is currently a single like-new ASUS ROG Flow Z13 laptop available on Amazon for $2,819.99:

https://www.amazon.com/gp/product/B0DW238TXK?smid=A2L77EE7U53NWQ&th=1

That's the lowest price I've seen all year, worth jumping on if you're looking to self host any model bigger than Qwen 3.8 27b. I like those laptops for mobile inference. I'm not aware of any other option which provides as much VRAM in a such a small package, for anywhere near the price.

Apple laptops with 128Gb VRAM start at $4099, for the model with the lowest number of cores, and their prefill speed is even slower than the Strix Halo (though generation speed can be faster): https://www.apple.com/shop/buy-mac/macbook-pro/14-inch-space-black-standard-display-apple-m5-max-chip-18-core-cpu-32-core-gpu-36gb-memory-2tb-storage

History

Nick AntonaccioAdmin
Sep 22, 2026 at 02:42

#16

And today, it's $2,759.99!

Please login to post a reply.