ai.canshu.com
中文
VRAM Calculator Advisor & TCO API Pricing Compare Models Leaderboard GPUs

Search models, GPUs, and tools

↑↓ Navigate · Enter Open · ★ Watchlist prioritized ai.canshu.com
Q4_K_M: Smooth Fit 77.26 GB / 108.8 GB Usable

Qwen3.5-122B-A10B (MoE) on NVIDIA DGX Spark GB10 (128GB Unified): VRAM Requirement & Speed Estimate

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.5-122B-A10B (MoE) requires 77.26 GB of VRAM (73.96 GB weights + 0.52 GB KV cache + 2.78 GB overhead), fitting within DGX Spark 128G's 108.8 GB usable memory at ~28.5 tokens/s.

Concurrency (Batch):
MoE 122B / A10B·48L (d=3072)·32Q/2KV·Ctx 256K·Text+Image+Video
Runnable Models →
128 GB Unified·273 GB/s·Safe Usable 108.8 GB
Smooth Fit77.26 / 108.8 GB~28.5 tok/s
3. Weight Quantization
4. Per-Request Context16K (16,384 tok)
Verified: 2026-11-20·Roofline bandwidth/compute estimate (Batch=1)
Smooth Fit
77.26/ 108.8 GB usable

Physical 128 GB · Load 71%

Theoretical Decode Speed
~28.5tokens/s

MoE: reads only 6.06 GB active weights/step

Max Context at Batch=1
256K/ 256K

Enable Q8/Q4 KV quantization to scale batch concurrency

Weights: 73.96 GBKV Cache (1×16K): 0.52 GBOverhead: 2.78 GB
Reserves ~15% unified memory for OS
🏃 Smooth & Snappy~17.5s for a 500-token response
Reference Launch CLI:
OLLAMA_CONTEXT_LENGTH=16384 ollama run qwen3.5:122b-a10b

DGX Spark 128G Fits Qwen3.5-122B-A10B (MoE) · Cost & TCO

Approx. $3,999 (~$31/GB VRAM) · 170W TDP

📊 All Quantization Levels for Qwen3.5-122B-A10B (MoE) on DGX Spark 128G

Quantization Total VRAM (16K) Verdict Est. Speed Max Context
FP16 / BF16 250.7 GB Out of VRAM (OOM) — —
FP8 / Q8_0 134.04 GB Out of VRAM (OOM) — —
GGUF Q6_K 103.86 GB Tight Fit (Watch Context) ~19.6 tok/s 212K
GGUF Q5_K_M 90.33 GB Smooth Fit ~24.4 tok/s 256K
Q4_K_M / AWQ 77.26 GB Smooth Fit ~28.5 tok/s 256K
GGUF IQ4_XS 67.93 GB Smooth Fit ~32.3 tok/s 256K
GGUF Q3_K_M 62.64 GB Smooth Fit ~35 tok/s 256K
GGUF Q2_K 50.82 GB Smooth Fit ~42.9 tok/s 256K

💡 Recommended Alternative GPUs & Macs for Qwen3.5-122B-A10B (MoE)

Frequently Asked Questions

Can NVIDIA DGX Spark GB10 (128GB Unified) run Qwen3.5-122B-A10B (MoE) locally?

Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.5-122B-A10B (MoE) requires 77.26 GB of VRAM (73.96 GB weights + 0.52 GB KV cache + 2.78 GB overhead), fitting within DGX Spark 128G's 108.8 GB usable memory at ~28.5 tokens/s.

What is the best quantization for Qwen3.5-122B-A10B (MoE) on DGX Spark 128G?

The highest-quality quantization that fits cleanly on DGX Spark 128G is GGUF Q6_K (Ultra High Quality), using 103.86 GB VRAM at ~19.6 tokens/s with up to 212K context.

How much VRAM does Qwen3.5-122B-A10B (MoE) need in full FP16 precision?

In unquantized FP16/BF16 precision (16K context), Qwen3.5-122B-A10B (MoE) requires 250.7 GB VRAM (244 GB for weights alone). Because it uses a Mixture-of-Experts (MoE) architecture, all 122B parameters must sit in memory, while only 10B active parameters are read per token.