LLM Roofline
AMD / Beelink / GMKtec / Minisforum · UMA · Backend: llama.cpp

Ryzen AI Max+ 395 Strix Halo (128 GB) — LLM Sizing & Speed Matrix

256-bit LPDDR5X-8000 UMA Mini-PC (GMKtec EVO-X2, Beelink GTR9 Pro, Minisforum MS-S1 Max). Up to 110 GiB VRAM via Linux TTM.

Reference Pricing
$1,899 Street
Usable Memory
96 GiB (110 tuned)
Memory Bandwidth
256 GB/s (74% eff.)
FP16 Compute
59.4 TFLOPS
Roofline Ridge
156.8 FLOP/B
Power (Load / Idle)
140W / 14W

Buy or Rent Ryzen AI Max+ 395 Strix Halo (128 GB)

Vendor & Cloud Links
GMKtec EVO-X2 (Strix Halo 128 GB)
GMKtec · Direct Hardware
↗
Beelink Strix Halo AI Mini-PC
Beelink · Direct Hardware
↗
Minisforum AI Workstation Mini-PC
Minisforum · Direct Hardware
↗

All 32 Open-Weight Models on Ryzen AI Max+ 395 Strix Halo (128 GB) (Q4_K_M @ 32k)

Model Status Weights Total @ 32k Decode TTFT (4k) Max Safe Ctx
Command A 111B COMFORTABLE 63.9 GiB 68.9 GiB 2.6 tok/s 31.54s 256k
Llama 4 Scout 109B-A17B COMFORTABLE 62.8 GiB 70.0 GiB 9.8 tok/s 5.00s 245k
Qwen2.5 72B Instruct COMFORTABLE 41.9 GiB 53.5 GiB 3.4 tok/s 20.83s 32k
DeepSeek-R1 Distill Llama 70B COMFORTABLE 40.7 GiB 52.3 GiB 3.5 tok/s 20.25s 128k
Llama 3.3 70B Instruct COMFORTABLE 40.7 GiB 52.3 GiB 3.5 tok/s 20.25s 128k
Llama 3.1 70B Instruct COMFORTABLE 40.7 GiB 52.3 GiB 3.5 tok/s 20.25s 128k
Llama-3.3 Nemotron Super 49B COMFORTABLE 28.7 GiB 36.5 GiB 5.1 tok/s 14.25s 128k
Mixtral 8x7B Instruct COMFORTABLE 26.9 GiB 32.0 GiB 13.6 tok/s 3.74s 32k
Qwen3 32B COMFORTABLE 18.9 GiB 28.1 GiB 6.6 tok/s 9.67s 40k
Qwen2.5 Coder 32B COMFORTABLE 18.9 GiB 28.1 GiB 6.6 tok/s 9.45s 32k
DeepSeek-R1 Distill Qwen 32B COMFORTABLE 18.9 GiB 28.1 GiB 6.6 tok/s 9.45s 128k
QwQ 32B Reasoning COMFORTABLE 18.9 GiB 28.1 GiB 6.6 tok/s 9.45s 40k
GLM-4 32B 0414 COMFORTABLE 18.6 GiB 21.9 GiB 8.6 tok/s 9.37s 32k
Qwen3 30B-A3B COMFORTABLE 17.6 GiB 21.4 GiB 31.7 tok/s 1.17s 40k
Gemma 3 27B IT COMFORTABLE 15.8 GiB 20.0 GiB 9.4 tok/s 7.88s 128k
Mistral Small 3.1 24B COMFORTABLE 13.8 GiB 20.1 GiB 9.4 tok/s 6.84s 128k
Codestral 22B COMFORTABLE 12.8 GiB 21.2 GiB 8.9 tok/s 6.55s 32k
Qwen3 14B COMFORTABLE 8.5 GiB 14.8 GiB 13.1 tok/s 4.35s 40k
Qwen2.5 14B Instruct COMFORTABLE 8.5 GiB 15.8 GiB 12.2 tok/s 4.39s 32k
DeepSeek-R1 Distill Qwen 14B COMFORTABLE 8.5 GiB 15.8 GiB 12.2 tok/s 4.39s 128k
Phi-4 14B COMFORTABLE 8.5 GiB 12.8 GiB 15.2 tok/s 4.32s 16k
Gemma 3 12B IT COMFORTABLE 7.0 GiB 10.4 GiB 18.9 tok/s 3.62s 128k
Qwen3 8B COMFORTABLE 4.7 GiB 10.3 GiB 19.1 tok/s 2.46s 40k
Llama 3.1 8B Instruct COMFORTABLE 4.6 GiB 9.7 GiB 20.5 tok/s 2.40s 128k
Qwen2.5 7B Instruct COMFORTABLE 4.4 GiB 7.2 GiB 28.8 tok/s 2.25s 32k
Gemma 3 4B IT COMFORTABLE 2.5 GiB 4.2 GiB 52.9 tok/s 1.30s 128k
Mixtral 8x22B Instruct TIGHT 81.2 GiB 89.6 GiB 5.3 tok/s 11.18s 64k
DeepSeek-R1 671B OOM 386.4 GiB 390.1 GiB OOM — OOM
DeepSeek-V3 671B OOM 386.4 GiB 390.1 GiB OOM — OOM
Llama 3.1 405B Instruct OOM 233.8 GiB 252.3 GiB OOM — OOM
Llama 4 Maverick 400B-A17B OOM 231.5 GiB 238.8 GiB OOM — OOM
Qwen3 235B-A22B OOM 135.4 GiB 142.4 GiB OOM — OOM

Interactive Ryzen AI Max+ 395 Strix Halo (128 GB) Simulator

111B Dense · GQA
256 GB/s · 59.4 TFLOPS
3. Weight Quantization 4.85 bpw
4. KV Cache
32,768 tok
Compute Roofline
COMFORTABLE FIT Memory-Bandwidth Bound
Max Safe Ctx: 256k
Memory Allocation Footprint 68.91 / 96.00 GiB usable
Weights: 63.93 GiB KV Cache: 2.75 GiB Overhead: 2.24 GiB Headroom: 27.09 GiB
Decode Speed
2.6
tok/s (batch=1)
Prefill Speed
130
prompt tok/s
TTFT Latency
31.54s
for 4,096 tok
Active Params
111.0B
of 111B total
llama.cpp Server Command 
llama-server -hf CohereLabs/c4ai-command-a-03-2025:Q4_K_M -ngl 99 -fa on -c 32768 --port 8080
Context KV Cache Total VRAM Decode Status
4k 1.00 GiB 66.32 GiB 2.7 tok/s COMFORTABLE
8k 1.25 GiB 67.41 GiB 2.7 tok/s COMFORTABLE
16k 1.75 GiB 67.91 GiB 2.7 tok/s COMFORTABLE
32k 2.75 GiB 68.91 GiB 2.6 tok/s COMFORTABLE
64k 4.75 GiB 70.91 GiB 2.6 tok/s COMFORTABLE
128k 8.75 GiB 74.91 GiB 2.4 tok/s COMFORTABLE

Buy-vs-Cloud-vs-API Monthly TCO Break-Even

BUY LOCAL
Own Hardware/mo
$57.96
1679.2h active + power
Cloud GPU/mo
$2684.13
@$1.39/hr spot/pod
Serverless API/mo
$275.00
Payback: 5.6 mo

Recommended Hardware & Cloud Deployment

Contextual to fit status

Hardware FAQ — Ryzen AI Max+ 395 Strix Halo (128 GB)

How much usable VRAM / GPU memory does Ryzen AI Max+ 395 Strix Halo (128 GB) actually provide for LLMs?

Out of the box, Ryzen AI Max+ 395 Strix Halo (128 GB) safely reserves 96 GiB for GPU inference, expandable up to 110 GiB with OS UMA tuning.

What is the largest Q4_K_M model Ryzen AI Max+ 395 Strix Halo (128 GB) can run at 32k context?

With Q4_K_M weights and FP16 KV cache at 32k context, the largest fitting model in our verified catalog on Ryzen AI Max+ 395 Strix Halo (128 GB) is Command A 111B (111B total).