Key Takeaways
- The SER8 runs Home Assistant, Frigate, Whisper, and a 7B local LLM on one box for about $650
- Ryzen 7 8845HS: 8 cores/16 threads, Radeon 780M iGPU, 32 GB DDR5 on two user-replaceable SO-DIMMs
- Local-LLM speed comes from the 780M iGPU and DDR5 bandwidth — a 7B model is comfortable
- Cheaper and simpler than a GEEKOM A9 Max; far more capable than an Intel N150 box
- Wi-Fi is 6 (not 6E) and there is one 2.5GbE port — fine for most home servers
- Made in China — factor 2026 US/EU import measures into landed cost (see trade note)
Verdict — Who Should Buy It
Buy the Beelink SER8 if you want the cheapest box that still runs a 7B local model at usable speed alongside Home Assistant. For most people building a local-AI smart home it is the right default: a strong Radeon 780M iGPU, 32 GB of DDR5, and dual replaceable RAM slots at about half the price of a GEEKOM A9 Max.
Its single strongest use case is an all-in-one home server — Home Assistant, Ollama, Whisper, and Frigate on one quiet machine. Step up only if you need a larger model (more RAM headroom) or an external-GPU path; step down to an Intel N150 box only if a tiny model is enough.
Specifications
All specs below were verified against Beelink and independent review sources in July 2026. RAM as sold varies by SKU (24, 32, or 64 GB); this table reflects the mainstream 32 GB / 1 TB unit.
- No discrete GPU: all AI inference runs on the CPU and Radeon 780M iGPU sharing system RAM.
- Dual SO-DIMM slots are user-replaceable; the practical DDR5 ceiling on this platform is around 96 GB.
| Spec | Beelink SER8 (8845HS) |
|---|---|
| CPU | AMD Ryzen 7 8845HS (Zen 4, 4 nm) |
| Cores / threads | 8 cores / 16 threads, up to 5.1 GHz |
| iGPU | AMD Radeon 780M — 12 CU, RDNA 3 |
| NPU | Ryzen AI, 16 TOPS (XDNA gen 1) |
| RAM | 32 GB DDR5-5600 (2×16 GB); dual SO-DIMM; user-replaceable (SKUs up to 64 GB) |
| Storage | 1 TB NVMe; 2× M.2 2280 PCIe 4.0; up to 8 TB |
| Ports | 1× USB4 (40 Gbps), 3× USB-A 10 Gbps, 2× USB-A 2.0, 1× USB-C 10 Gbps, HDMI 2.1, DP 1.4, 3.5 mm |
| Networking | 1× 2.5GbE, Wi-Fi 6 (not 6E), Bluetooth 5.2 |
| TDP | 54–65 W (runs at 65 W sustained) |
| Dimensions | 135 × 135 × 44.7 mm, ~0.75 kg |
| OS | Windows 11 Pro (Linux/Proxmox install supported) |
| Price (indicative) | ~$650 (July 2026 — volatile, check current price) |
Local AI & LLM Performance
With 32 GB of RAM the SER8 comfortably runs a 7B model on Ollama and can load a 13B–14B model at 4-bit, though larger models are limited by memory bandwidth rather than capacity — all figures estimated from the verified hardware, not a measured benchmark. The Radeon 780M is the key part: on dual-channel DDR5-5600 it keeps a 7B assistant responsive next to Home Assistant.
- Ollama, LM Studio, and llama.cpp run well; the 780M is used through Vulkan or ROCm backends.
- A 7B model is the comfortable sweet spot; 13B–14B works for non-interactive use but feels slower.
- 32B-class models fit only tightly in 32 GB and run slowly — step up to a 64 GB SKU for those.
- The 16-TOPS NPU is not used by mainstream LLM runtimes; the iGPU and RAM bandwidth set latency.
- See Ollama on Home Assistant to connect the model, and the local LLM hardware guide for VRAM depth.
Pros & Cons
The SER8 is the value benchmark; its compromises are minor for a home server. Balanced view below.
- Pros
- Best value for a capable local-AI box — ~$650 for a 7B-ready machine
- Strong Radeon 780M iGPU and 32 GB DDR5 handle Home Assistant plus a 7B model
- Dual user-replaceable SO-DIMM slots and two M.2 slots — upgradeable
- USB4, HDMI 2.1, and DP 1.4 support triple 4K output
- Runs at 65 W — modest for an always-on server
- Cons
- Wi-Fi is 6, not 6E — use the 2.5GbE port for a reliable wired link
- Only one 2.5GbE port (rivals like the UM890 Pro offer two)
- Not enough RAM at 32 GB for 30B-plus models without an upgrade
- The NPU does not accelerate local LLMs
Buyer Context: Tariffs & Availability (2026)
The SER8 is manufactured in China, so its landed price reflects 2026 trade measures — a reason to treat any price as a moving target. The facts below are buyer context, not editorial opinion.
- US: Section 301 tariffs on Chinese electronics remain in force in 2026; the separate 2025 "IEEPA" tariffs were struck down by the Supreme Court in February 2026 and replaced by a temporary, capped Section 122 tariff. The sub-$800 duty-free "de minimis" exemption for direct imports has also ended.
- EU (affects DE/FR): there is no broad EU tariff on finished mini PCs, but from July 2026 the €150 duty-free threshold on low-value direct-from-China parcels was removed and a small per-parcel handling fee added.
- Assumption (macro trend, not a per-product fact): redirected Chinese export capacity has kept availability of these brands high in the EU and US, which broadly supports competitive pricing.
- Net effect: verify the current price at the retailer before buying — the ~$650 figure here is indicative and date-stamped July 2026.
Where to Buy & Current Price
Prices move week to week, so check the live price rather than trusting a fixed figure. As of July 2026 the 32 GB / 1 TB SER8 clusters around $650, with the official store often higher than street listings.
- Confirm the RAM SKU (24, 32, or 64 GB) — the 32 GB unit is the value sweet spot for a 7B model.
Alternatives to Consider
If the SER8 is not the right fit, four boxes bracket it on price and capability.
- GEEKOM A9 Max — the headroom pick: 128 GB RAM ceiling and Wi-Fi 7, for roughly double the price
- Minisforum UM890 Pro — similar Radeon 780M, but adds dual 2.5GbE, Wi-Fi 6E, and an OCuLink eGPU port
- Beelink EQ14 — budget Intel N150 box for Home Assistant plus a tiny model
- GMKtec G3 Plus — another budget N150 option with upgradeable RAM
- Still comparing? Start from the best mini PCs for Home Assistant + local AI roundup.
Frequently Asked Questions
Can the Beelink SER8 run local LLMs?
Yes. With 32 GB of RAM and the Radeon 780M iGPU it comfortably runs a 7B model on Ollama and can load a 13B–14B model at 4-bit. This is estimated from its verified hardware rather than a fixed benchmark, because speed depends on the model, quantization, and backend.
How big a model fits in its RAM?
At 32 GB, a 7B model is comfortable and a 13B–14B model at 4-bit is usable but slower. A 30B-class model fits only tightly and runs slowly; for those, choose the 64 GB SKU or a box with a higher RAM ceiling.
Is the SER8 good for Home Assistant and always-on use?
Yes. It runs Home Assistant, Ollama, Whisper, and Frigate together on 32 GB and draws about 65 W under load, so continuous operation is inexpensive. Use the 2.5GbE port for a reliable wired connection.
Is the RAM upgradeable?
Yes. The SER8 uses two standard DDR5 SO-DIMM slots that are user-replaceable, so you can start at 32 GB and move to 64 GB later. It also has two M.2 slots for storage expansion.
Does the SER8 have Wi-Fi 6E?
No. The SER8 ships with Wi-Fi 6 (an Intel AX200 module), not Wi-Fi 6E. For a home server this rarely matters — a wired 2.5GbE link is more reliable than any Wi-Fi band.
How much power does the SER8 use?
It runs at up to 65 W under sustained load and far less at idle, so leaving it on continuously is cheap. The exact figure depends on the workload and the power profile you choose.
SER8 or GEEKOM A9 Max?
The SER8 is the value choice and covers most local-AI smart-home needs for about $650. Choose the A9 Max only if you need its 128 GB RAM ceiling, Wi-Fi 7, or extra vision headroom, which roughly doubles the cost.
Where is the Beelink SER8 made, and does that affect price?
It is manufactured in China. In 2026 US import measures (Section 301 tariffs, the end of the sub-$800 de minimis exemption) and the EU removal of the €150 low-value parcel exemption can affect landed cost, so check the current retailer price rather than relying on a fixed number.