Aoostar Overseas Fulfillment Store via AliExpress has
AOOSTAR MACO Barebones Mini PC (No SSD) on sale for $308.14 - $42 with promo code
MYUSDEAL42 at checkout =
$266.14.
Shipping is free.
Thanks to Community Member
thethingx for sharing this deal.
- Note: Price may fluctuate slightly with currency exchange rates. You must be signed into your AliExpress account and have your address added to add items to your cart and place an order. You may have to manually type in the promo code to apply it at checkout. This item ships from a USA warehouse. For expedited customer service, please email [email protected].
Details:
- AMD Ryzen 7 Pro 6850H 8-cores, 16-threads (3.2GHz Base / 4.7GHz
- Radeon 680M Integrated Graphics
- 24GB LPDDR5 6400Mhz RAM (Soldered; cannot be removed or replaced)
- Wi-Fi 6 + Bluetooth 5.2
- Internal Storage Expansion:
- 3x M.2 2280
- 4x PCIe 4.0
- Max 24TB
- Ports:
- 4x USB-A
- 2x USB-C
- 1x OCuLink (Does not support hot-swap)
- 2x LAN (2.5G)
- 1x 3.5mm Audio Jack
- 1x DisplayPort 1.4
- 1x HDMI 2.1
Top Comments
Essentially put the steam deck at your baseline, and increase the CPU performance, and thats where you expect this to land.
Community Reviews
This barebones Ryzen 7 Pro mini PC stands out as a strong value for buyers who already have an NVMe drive, thanks to 24GB of fast LPDDR5 memory, USB4, and OCuLink support. The 680M graphics and 8 core Zen 3 CPU are described as well above Steam Deck class performance, with enough output for emulation, Plex, and even four displays. The main catch is the missing SSD, and it is not a great fit for AI token generation workloads because there is no NPU and software support for the GPU is limited. A capable compact system if you can supply storage and do not need heavy AI acceleration.
Arcade cabinet builders wanting a compact emulation box, because the Ryzen 7 Pro chip and 12-core Radeon graphics should handle game emulation well.
PC tinkerers who already have a spare SSD, because the barebones unit includes fast LPDDR5 memory and OCuLink plus dual USB4 for expansion.
Home theater users doing Plex transcoding, because the 8-core Zen 3 CPU is a stronger fit than the slower chips at this price.
AI agent builders needing local token generation, because it lacks an NPU and the GPU has weak ROCm and Vulkan support.
People expecting an all-in-one desktop without extra parts, because it ships without an SSD.
Power users who want Intel-style efficiency for always-on workloads, because it will not be as energy efficient as a Quick Sync Intel CPU.
Common Questions (4)
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!
36 Comments
Sign up for a Slickdeals account to remove this ad.
slightly better than steam deck, and the lenovo legion go S (8 core zen3 over 4 core zen2/zen3), 12CU rdna2 over the 8 in the steamdeck.
Essentially put the steam deck at your baseline, and increase the CPU performance, and thats where you expect this to land.
This machine should be like 35-45% faster GPU performance and like 70-80% faster in CPU performance. + it has 8GB more of similar LPDDR5 RAM?
I wouldn't call that "slightly" better.
EDIT: on top of that, this has an oculink port for use with an eGPU. AND it has 2 USB4s. Can't do that on the steam deck.
If you have a spare SSD, this is a sure bet. It isnt Zen4, but for THIS configuration, its actually cheaper than the equivalent (with even less RAM) from summer 2025.
Sign up for a Slickdeals account to remove this ad.
But if any of those tasks require any sort of token generation both input and output (as opposed to just initiating a command or scraping for information), you will need more power (and more VRAM), and this machine is a nonstarter for that. There is no NPU, and the GPU has middling support, even with vulcan. Most backends do not recognize the 680m with ROCm, and Vulcan usage is VERY slow on these.
To bypass this issue you can rent GPU access on openrouter and get an API key, or use a pioneer cloud model like GPT 5.5. (preferably one that does not bill per token), as well. Just dont use claude, because they will consider any of these AI agents as "additional usage" that is not billed against your sub, but as an additional charge.
You can technically use the oculink port and attach a decent GPU to do this task as well...but then the whole SD aspect of this deal is done. Once you spend enough, its just cheaper to get a mac mini.
Sign up for a Slickdeals account to remove this ad.
it depends on how you define "run." As an AI agent, it is more than powerful enough to delegate tasks. You can even use an RPI for this.
But if any of those tasks require any sort of token generation both input and output (as opposed to just initiating a command or scraping for information), you will need more power (and more VRAM), and this machine is a nonstarter for that. There is no NPU, and the GPU has middling support, even with vulcan. Most backends do not recognize the 680m with ROCm, and Vulcan usage is VERY slow on these.
To bypass this issue you can rent GPU access on openrouter and get an API key, or use a pioneer cloud model like GPT 5.5. (preferably one that does not bill per token), as well. Just dont use claude, because they will consider any of these AI agents as "additional usage" that is not billed against your sub, but as an additional charge.
You can technically use the oculink port and attach a decent GPU to do this task as well...but then the whole SD aspect of this deal is done. Once you spend enough, its just cheaper to get a mac mini.
Join The Conversation
Share your experience with the Slickdeals community
Share information with the community. Please follow our Community Guidelines and be kind!