Best Buy has
ASUS ROG Flow Z13 Touchscreen 2-in-1 Gaming Laptop (Open Box - Excellent, GZ302EA-R9641TB) for
$1757.99.
Shipping is free or select free store pickup where available.
Thanks to Community Member
Dr.W for sharing this deal.
Note: Availability for store pickup will vary by location.
Also Available (store pickup only):
Specs:
- 13.4" 2560x1600 WQXGA 180Hz IPS-Level Touchscreen Display
- AMD Ryzen AI MAX+ 395 16-Core Processor
- AMD Radeon 8060S Graphics
- AMD XDNA NPU up to 50TOPS
- 64GB LPDDR5X 8000 Memory (Onboard)
- 1TB M.2 2230 PCIe 4.0 NVMe Solid State Drive
- Backlit RGB Chiclet Keyboard
- Wi-Fi 7(802.11be) (Triple band) 2x2 + Bluetooth 5.4
- Windows 11 Home
- Ports:
- 1x 3.5mm Combo Audio Jack
- 1x HDMI 2.1 FRL
- 2x Type-C USB 4 with support for DisplayPort / power delivery (data speed up to 40Gbps)
- 1x USB 3.2 Gen 2 Type-A (data speed up to 10Gbps)
- 1x card reader (microSD) (UHS-II)
- 70WHr Battery
- ~2.65 lbs
Top Comments
Ridiculously powerful, lots of use cases, replaced my entire desktop setup with it (TB4 Dock when at home, connected to a 49" Samsung).
Do yourself a massive favor and get G-Helper, don't bother with Asus' bloatware (Armoury Crate and MyAsus). Gives all the necessary features in a tiny and lightweight package, low overhead, and even more controls for undervolting, overclocking, etc. I've got modes setup for completely silent, balanced, and then an overdrive setting.
These price points are a steal, especially considering the RAM scenario right now.
Community Reviews
This open box ASUS ROG Flow Z13 stands out as a compact, very powerful Windows tablet for local AI work and gaming. The 64GB memory and Ryzen AI Max Plus 395 are praised for running large local models, including Qwen 3.6 35B and other 70B class setups, while gaming stays smooth even in recent titles and on lower power settings. Battery life can stretch through long flights with a 70W charger, and Linux support plus SSD upgrades have been reported as straightforward. The main tradeoffs are the 13.4 inch screen, open box uncertainty, and possible warranty reduction if the previous owner registered it.
Common Questions (5)
Yes, 64GB is enough for local text models and can handle around 70B models in Q4, with some users running Qwen 3.6 35B on it. It is slower than a desktop RTX 5090 or higher-memory remote boxes, but it can fit larger models and larger context windows than smaller-memory laptops.
Yes. The open-box units are straight returns that were not checked when dropped off, not fake refurbished stock. That means condition can vary, so it is a bit of a lottery compared with buying new.
It can charge from lower-wattage USB-C chargers, but some chargers cause the battery to cycle between charging and discharging because the USB-C port does not pass power through. A custom power profile between Quiet and Balanced can reduce that on/off cycling and make lower-wattage charging more usable.
The SSD is the only reported upgradeable component, and a 2TB drive has been installed successfully. Fedora has been used on it with no issues.
If the previous owner registered the device, ASUS may give it a modified 3-month warranty instead of the full year when you register it. That makes warranty status worth checking before buying open-box.
AI-generated summary based on comments from the community. Updated ago
Join The Conversation
Share your experience with the Slickdeals community
Share information with the community. Please follow our Community Guidelines and be kind!
33 Comments
Sign up for a Slickdeals account to remove this ad.
For $1600 for 64GB version that is hella steal. I paid almost $2400 few months ago
Our community has rated this post as helpful. If you agree, why not thank
Ridiculously powerful, lots of use cases, replaced my entire desktop setup with it (TB4 Dock when at home, connected to a 49" Samsung).
Do yourself a massive favor and get G-Helper, don't bother with Asus' bloatware (Armoury Crate and MyAsus). Gives all the necessary features in a tiny and lightweight package, low overhead, and even more controls for undervolting, overclocking, etc. I've got modes setup for completely silent, balanced, and then an overdrive setting.
These price points are a steal, especially considering the RAM scenario right now.
Ridiculously powerful, lots of use cases, replaced my entire desktop setup with it (TB4 Dock when at home, connected to a 49" Samsung).
Do yourself a massive favor and get G-Helper, don't bother with Asus' bloatware (Armoury Crate and MyAsus). Gives all the necessary features in a tiny and lightweight package, low overhead, and even more controls for undervolting, overclocking, etc. I've got modes setup for completely silent, balanced, and then an overdrive setting.
These price points are a steal, especially considering the RAM scenario right now.
For $1600 for 64GB version that is hella steal. I paid almost $2400 few months ago
128GB? OK, I can run a model in the 60-80GB range with context. Huge step up from the smaller models.
But 64GB of RAM? ...
Better to have a headless machine over the network with 128 GB accessed remotely. It's worth the step up.
Or the 128gb of this, but then are you truly going to run your llm on a laptop? Again, may as well just ssh or tailscale into a remote box with more ram to work with, strix halo, dgx spark, or sufficiently sized pcie cards.
Sign up for a Slickdeals account to remove this ad.
128GB? OK, I can run a model in the 60-80GB range with context. Huge step up from the smaller models.
But 64GB of RAM? ...
Better to have a headless machine over the network with 128 GB accessed remotely. It's worth the step up.
Or the 128gb of this, but then are you truly going to run your llm on a laptop? Again, may as well just ssh or tailscale into a remote box with more ram to work with, strix halo, dgx spark, or sufficiently sized pcie cards.
128GB? OK, I can run a model in the 60-80GB range with context. Huge step up from the smaller models.
But 64GB of RAM? ...
Better to have a headless machine over the network with 128 GB accessed remotely. It's worth the step up.
Or the 128gb of this, but then are you truly going to run your llm on a laptop? Again, may as well just ssh or tailscale into a remote box with more ram to work with, strix halo, dgx spark, or sufficiently sized pcie cards.
64GB is nice though, that's worth what $800 now? lol. crazy ram prices.
I use M5 Max 128GB because it's more functional (personal ref on apple eco) and much faster an any AI Max mchine.
I do run text based LLM on my Macboook though (nothing larger than 35B for now--- much slower than Nvidia running the same model but I run it there cuz I save nvidia for more intensive work via tailscale -- this is the way and you definitely have a great tip for everyone.
tailscale is a savior for me!!!
sometimes I use a Neo and tailscale to all my machines at home.
Based on my very little research, these are cheaper but much slower in inferencing.
64GB is nice though, that's worth what $800 now? lol. crazy ram prices.
I personally would buy M5 Max 128GB (and I did) over Any one of these Ryzen AI 128GB because M5 Max will run AI much faster.
Nvidia still king at speed as long as you fit it into the VRAM on the GPUs though. I think this is a poor man's LLM machine if you want it for AI but it is a lot slower if I recall and it will get annoying.
I said little at first cuz I assumed someone will write a more accurate real take of their experience along the way.
Based on my research before I bought some of my gear. These are cheaper but much slower in inferencing.
64GB is nice though, that's worth what $800 now? lol. crazy ram prices.
I use M5 Max 128GB because it's more functional (personal ref on apple eco) and much faster an any AI Max mchine.
I do run text based LLM on my Macboook though (nothing larger than 35B for now--- much slower than Nvidia running the same model but I run it there cuz I save nvidia for more intensive work via tailscale -- this is the way and you definitely have a great tip for everyone.
tailscale is a savior for me!!!
sometimes I use a Neo and tailscale to all my machines at home.
So, it is not that this AMD processor lacks in AI computing power compare to M5 Max, it lacks memory speed by a mile...
Sign up for a Slickdeals account to remove this ad.
Join The Conversation
Share your experience with the Slickdeals community
Share information with the community. Please follow our Community Guidelines and be kind!