Can NVIDIA DGX Spark 128GB run GLM-4.5-Air?
Estimated speed and memory fit for GLM-4.5-Air (110.47B MoE, 12B active) on NVIDIA DGX Spark 128GB. We have no measured run for this pair yet, so the speed is an estimate; submit yours.
Short answer
Yes. GLM-4.5-Air Q4_K_M fits fully on NVIDIA DGX Spark 128GB at 16k context, using 76.1 of 108.3 GB usable, and generates about ~17 tokens/s (estimated, not measured).
Measured generation speed by quantization
No measured rows for GLM-4.5-Air on NVIDIA DGX Spark 128GB yet. The ~17 tok/s above is a bandwidth estimate (273 GB/s × 0.81 efficiency). Have this machine? Submit a llama-bench result and this page will switch to measured data.
Generation = tg128 (128 generated tokens, short context). Public rows are medians of community llama.cpp results (sources); our own runs use llama-bench on a rented GPU.
Alternatives
Less memory that still fits GLM-4.5-Air
Smaller memory usually means a cheaper card or Mac (we do not list prices).
- Mac Studio M5 Max 128GB: Q5_K_M, 89 GB usable, ~33 tok/s estimated
- MacBook Pro 14/16 M5 Max 128GB: Q5_K_M, 89 GB usable, ~33 tok/s estimated
- MacBook Pro 14/16 (2024) M4 Max 128GB: Q5_K_M, 89 GB usable, ~22 tok/s estimated
Faster for GLM-4.5-Air
- RTX PRO 6000 96GB: Q5_K_M, 95 GB usable, ~121 tok/s scaled from measured
- Mac Studio M5 Ultra 256GB: Q8_0, 179 GB usable, ~46 tok/s estimated
- Mac Studio M5 Ultra 512GB (ships late Oct 2026): Q8_0, 358 GB usable, ~46 tok/s estimated
Other models on NVIDIA DGX Spark 128GB, similar size
- gpt-oss 120B: Q8_0, fits fully, ~57 tok/s (scaled from measured)
- Nemotron 3.5 Super VL 120B-A12B (hardware page): Q4_K_M, fits fully, ~21 tok/s (scaled from measured)
- Llama 3.3 70B (hardware page): Q4_K_M, fits fully, ~5 tok/s (estimated)
- Qwen3.6 35B-A3B (hardware page): Q8_0, fits fully, ~49 tok/s (measured)
Memory and context limits
Q4_K_M weights are 73.0 GB; the 16k f16 KV cache adds 3.09 GB. Largest context that fits in 108.3 GB usable:
| KV cache type | Max context |
|---|---|
| f16 | 128k |
| q8_0 | 128k |
| q4_0 | 128k |
Native limit 128k. Full KV cache table for GLM-4.5-Air.
Copy-paste commands
The advisor's pick for this hardware: Q4_K_M, 16k context.
llama-server -hf unsloth/GLM-4.5-Air-GGUF:Q4_K_M -c 16384 -fa on --jinja -ngl 99
OLLAMA_CONTEXT_LENGTH=16384 ollama serve ollama run hf.co/unsloth/GLM-4.5-Air-GGUF:Q4_K_M
FAQ
How fast is GLM-4.5-Air on NVIDIA DGX Spark 128GB?
About 17.1 tokens/s for generation (Q4_K_M, estimated from 273 GB/s memory bandwidth). Speed drops as the context fills; see the depth table on this hardware's other pages.
How much memory does GLM-4.5-Air need on NVIDIA DGX Spark 128GB?
The advisor's pick, Q4_K_M, is 73.0 GB of weights plus 3.09 GB of KV cache at 16k context (76.1 GB), against 108.3 GB usable.
Is the GLM-4.5-Air speed on NVIDIA DGX Spark 128GB measured or estimated?
Estimated. No measured llama.cpp run for this pair is on file yet, so the figure comes from memory bandwidth and our calibration against measured hardware.
What can I use instead of NVIDIA DGX Spark 128GB for GLM-4.5-Air?
With less memory, Mac Studio M5 Max 128GB (~33 tok/s), MacBook Pro 14/16 M5 Max 128GB (~33 tok/s), MacBook Pro 14/16 (2024) M4 Max 128GB (~22 tok/s) also fit it fully. Faster: RTX PRO 6000 96GB (~121 tok/s), Mac Studio M5 Ultra 256GB (~46 tok/s), Mac Studio M5 Ultra 512GB (ships late Oct 2026) (~46 tok/s).
What is the maximum context for GLM-4.5-Air on NVIDIA DGX Spark 128GB?
With an f16 KV cache about 128k tokens, 128k with q8_0 and 128k with q4_0, bounded by the model's own limit.
Related
All models on NVIDIA DGX Spark 128GB · GLM-4.5-Air on all hardware · VRAM tier · all measured pages · leaderboard
Compare: AMD Ryzen AI Max+ 395 (Strix Halo) 128GB vs NVIDIA DGX Spark 128GB