Model × hardware check
Can GLM-5.3-Flash run on AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB?
Yes, this is a comfortable estimate at 8,192 tokens. a 320B-total, 18B-active multimodal MoE with hybrid sparse and linear attention The Strix Halo mini PC provides 128 GB of shared memory, though firmware and OS allocation can leave substantially less usable by the GPU.
Short answer
Comfortable fit
At least 20% memory headroom remains after weights, KV cache, and runtime overhead.
- Available shared CPU/GPU memory
- 128 GB
- Estimated requirement
- 105.3 GB
- Headroom
- +22.7 GB
- Fit ratio
- 122%
- Recommended quantization
- UD-IQ2_XXS
- Context ceiling
- 232,077 tokens
- Speed estimate
- Estimate · 24.3–33.1 tok/s
Estimate · decode · llama.cpp/MLX · 8,192-token context This is a bandwidth model, not a measured benchmark.
What the numbers say
The selected GLM-5.3-Flash package uses 103.0 GB of weight files. The 8,192 tokens KV cache estimate is 0.8 GB, and the formula adds 1.5 GB of runtime overhead. That produces a 105.3 GB requirement against 128 GB of shared CPU/GPU memory.
usable GPU memory ~96-112GB depending on BIOS/OS; also 64GB variant This is a compact shared-memory system, not a dedicated-VRAM GPU. The audit warns that usable GPU memory may be roughly 96-112 GB depending on BIOS, OS, and allocation settings, so the full 128 GB should not be treated as universally available to one model.
The context ceiling shown above is the point where the same formula reaches the machine's capacity. It is not a promise that the backend can allocate that full context after the operating system and other applications take their share.
Quantization trade-offs at 8,192 tokens
| Quantization | Weight files | 8,192 tokens requirement | Fit on this machine | Context ceiling | Speed estimate | Ollama |
|---|---|---|---|---|---|---|
| UD-IQ1_M | 98.71 GB | 101.0 GB | Comfortable fit · 127% | 274,158 tokens | Estimate · 25.4–34.6 tok/s | No verified tag |
| UD-IQ2_XXS | 102.97 GB | 105.3 GB | Comfortable fit · 122% | 232,077 tokens | Estimate · 24.3–33.1 tok/s | No verified tag |
| UD-Q2_K_XL | 109.85 GB | 112.2 GB | Tight fit · 114% | 164,260 tokens | Estimate · 22.8–31.1 tok/s | No verified tag |
| UD-Q3_K_XL | 148.66 GB | 151.0 GB | Won't fit · 85% | Does not fit | Not applicable — does not fit | No verified tag |
| UD-Q4_K_XL | 200.84 GB | 203.2 GB | Won't fit · 63% | Does not fit | Not applicable — does not fit | No verified tag |
| Q8_0 | 342.11 GB | 344.4 GB | Won't fit · 37% | Does not fit | Not applicable — does not fit | No verified tag |
For MoE models, a tight CPU-offload result means the weights fit in system memory even when they do not fit in accelerator memory. That does not increase the accelerator context ceiling.
Run commands
llama-cli -hf unsloth/GLM-5.3-Flash-GGUF:UD-IQ2_XXS -c 8192 -ngl 99 -p "Your prompt"Commands are examples, not a guarantee that the selected quantization tag is available in every runtime. Confirm the package name and license on the linked model page.
Public speed checks
No verified measurement yet. The estimate above is deliberately labeled as an estimate rather than a benchmark.
If it does not fit
This pairing already clears the comfort threshold
The current estimate leaves +22.7 GB. A longer context or another GPU application can reduce that margin, so verify free memory before relying on it.
Low-cost setup upgrades
Useful upgrades around AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB
These are optional setup improvements, not a claim that an accessory can increase GPU or unified-memory capacity.
External NVMe SSD for model storage
Keeps large GGUF and MLX files off the boot drive. Fast storage reduces load time, but it does not increase the memory available for inference.
Search current listings on AmazonDesktop RAM upgrade kit
Useful for MoE CPU offload and keeping the operating system out of VRAM. Match the motherboard's memory type, slot layout, capacity limit, and validated part numbers.
Search current listings on AmazonGPU support bracket
A lower-cost mechanical upgrade for heavy three-slot cards. Check clearance, mounting height, and whether the bracket blocks fans or case panels.
Search current listings on AmazonAs an Amazon Associate I earn from qualifying purchases. Links are sponsored search links and do not change the price you pay. Affiliate disclosure.
Where this estimate can break
- The practical floor is about 100 GB of memory for the smallest 1-bit-class GGUF plus the projector, KV cache, and runtime workspace. A 128 GB machine can load it, but this is a low-bit, slow, and sensitive configuration rather than a normal 24 GB desktop workload.
- Ollama's glm-5.3-flash library entry is cloud-only. This site does not generate an ollama run command for it.
- A community 4-bit MLX conversion exists, but its card says mlx-lm does not support the glm5_next architecture and requires mlx-vlm. It also needs at least 256 GB of Mac memory, so it is not treated as a generic MLX runtime here.
- The 1M-token context is a capacity ceiling, not a practical default. Long contexts can still raise KV-cache and indexing memory beyond the 100 GB weight floor.
- Other GPU applications, display buffers, and the operating system reduce usable memory.
- Longer prompts, concurrent requests, and a different backend can raise peak memory.
Compare this pairing
FAQ
Questions about this pairing.
How much memory does GLM-5.3-Flash need on AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB?
At 8,192 tokens with UD-IQ2_XXS, the estimate is 105.3 GB: 103.0 GB of weights, 0.8 GB of KV cache, and 1.5 GB of runtime overhead. The available shared CPU/GPU memory is 128 GB.
What context length can AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB handle with this model?
The calculated ceiling is 232,077 tokens for this quantization, capped by the model's native 1,048,576-token context. Actual usable context can be lower after the OS, other applications, and backend workspace.
How does quantization change the memory requirement on AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB?
At 8,192 tokens, the published variants calculate as follows: UD-IQ1_M: 101.0 GB, comfortable fit; UD-IQ2_XXS: 105.3 GB, comfortable fit; UD-Q2_K_XL: 112.2 GB, tight fit; UD-Q3_K_XL: 151.0 GB, won't fit; UD-Q4_K_XL: 203.2 GB, won't fit; Q8_0: 344.4 GB, won't fit. A lower-bit file can move a result from too large to tight, but quality and kernel support may differ.
What can I run if GLM-5.3-Flash does not fit on AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB?
This selection already clears the comfort threshold. If you later increase the context, monitor the ceiling of 232,077 tokens; smaller alternatives include GLM-4.7-Flash, Flint 40B, NVIDIA Nemotron 3.5 Bolt E4B.
Is the speed estimate for GLM-5.3-Flash on AMD Ryzen AI Max+ 395 (Strix Halo) mini PC · 128 GB measured?
The site's Estimate · 24.3–33.1 tok/s range is an estimate from 18B active parameters, 256 GB/s accelerator memory bandwidth, and the selected quant at 8,192 tokens. It is not a benchmark. No public measurement is attached to this exact runtime and context.