MODEL ARCHITECTURE SPEC โข ALIBABA CLOUD35B MoE
Qwen 3.6 35B-A3B MoE VRAM & inference speed
35B Total Params3B Active (MoE)48 Layers โข 4 KV Heads32K Default Ctx (Max 128K)Ollama: qwen3.6:35b-a3b
Accurate memory requirements, KV cache footprint, and tokens/sec generation benchmarks for running Qwen 3.6 35B-A3B MoE across local GPUs, Apple Silicon Macs, and cloud instances.
โ
Runs 100% in GPU VRAM (Optimal Headroom)24.1 GB (45%)~78 t/s
Qwen 3.6 35B-A3B MoE on Apple Mac Mini M4 Pro (64GB Unified Memory)
โ Fits Comfortably in VRAM
โ FITS IN VRAM
Required VRAM
24.1 GB
45% of hardware capacity (54 GB)Est. Decode Speed
~78 tok/s
3B active params @ 273 GB/sAvailable Headroom
+29.9 GB
Safe for OS display & full contextMemory Allocation & Headroom
24.1 GB allocated of 54 GB capacity
45% utilized
Weights: 21 GB
KV: (32K): 1.61 GB
Overhead: 1.49 GB
Free Headroom: +29.9 GB
โก Architecture & Throughput Transparency: Single-user decode speed (~78 tok/s) is memory-bandwidth bound at batch size = 1 on Apple Mac Mini M4 Pro (64GB Unified Memory) (273 GB/s). Prompt prefill / TTFT (~1,131 tok/s) is compute-bound and utilizes GPU Tensor cores. Actual latency varies with FlashAttention-2 and runtime engine (Ollama/vLLM/llama.cpp).
OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=q8_0 ollama run qwen3.6:35b-a3bCloud Instance Options for Qwen 3.6 35B-A3B MoE
On-DemandDeploy this model on an on-demand cloud GPU without local hardware constraints or power usage:
RunPod Cloud GPUsBest 1-Click Ollama & vLLM
$0.34/hr (RTX 4070 Ti) โ $2.39/hr (H100)Vast.ai GPU MarketplaceCheapest Dual RTX 3090 / 4090
$0.18/hr (RTX 3060) โ $0.55/hr (RTX 4090)Lambda GPU CloudEnterprise 8x H100 Clusters
$1.99/hr (GH200 / H100)Quantization Precision Matrix
32K ctx โข KV: Q8_0| Quantization | Weights | KV Cache | Total VRAM | Est. Speed | Fit Status |
|---|---|---|---|---|---|
| FP1616 bpw | 70 GB | 1.61 GB | 74.81 GB (139%) | 0 t/s | OOM |
| Q8_08.5 bpw | 37.1 GB | 1.61 GB | 40.76 GB (75%) | ~48 t/s | FITS |
| Q6_K6.56 bpw | 28.7 GB | 1.61 GB | 32.06 GB (59%) | ~60 t/s | FITS |
| Q5_K_M5.68 bpw | 24.85 GB | 1.61 GB | 28.08 GB (52%) | ~68 t/s | FITS |
| Q4_K_M4.8 bpw | 21 GB | 1.61 GB | 24.1 GB (45%) | ~78 t/s | FITS |
| FP44.25 bpw | 18.55 GB | 1.61 GB | 21.56 GB (40%) | ~86 t/s | FITS |
| IQ3_XS3.3 bpw | 14.35 GB | 1.61 GB | 17.21 GB (32%) | ~105 t/s | FITS |
Export Full Architecture & Infrastructure Spec
Get an executive-ready PDF report with exact memory math, hardware benchmarks, and cost comparisons.
Hardware Compatibility Matrix for Qwen 3.6 35B-A3B MoE
Standard Q4_K_M precision with 32K context window and Q8_0 KV cache
| Hardware System | VRAM Envelope | Required VRAM | Headroom | Est. Decode Speed | Compatibility | Test Fit |
|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5090 (32GB GDDR7) Consumer GPU โข 1792 GB/s | 32 GB | 24.1 GB | +7.9 GB | ~553 tok/s | โ FITS | Test Fit โ |
| NVIDIA GeForce RTX 4090 (24GB GDDR6X) Consumer GPU โข 1008 GB/s | 24 GB | 24.1 GB | -0.1 GB | ~68 tok/s | โก PARTIAL | Test Fit โ |
| 2x NVIDIA RTX 3090 Dual Rig (48GB Total VRAM) Multi-GPU Rig โข 936 GB/s | 48 GB | 24.1 GB | +23.9 GB | ~263 tok/s | โ FITS | Test Fit โ |
| NVIDIA GeForce RTX 5080 (16GB GDDR7) Consumer GPU โข 960 GB/s | 16 GB | 24.1 GB | -8.1 GB | 0 (OOM) | โ OOM | Test Fit โ |
| NVIDIA RTX 4070 Ti Super (16GB) Consumer GPU โข 672 GB/s | 16 GB | 24.1 GB | -8.1 GB | 0 (OOM) | โ OOM | Test Fit โ |
| NVIDIA GeForce RTX 3060 (12GB) Consumer GPU โข 360 GB/s | 12 GB | 24.1 GB | -12.1 GB | 0 (OOM) | โ OOM | Test Fit โ |
| Apple Mac Mini M4 Pro (64GB Unified Memory) Apple Silicon โข 273 GB/s | 54 GB | 24.1 GB | +29.9 GB | ~78 tok/s | โ FITS | Test Fit โ |
| Apple Mac Studio M4 Max (128GB Unified Memory) Apple Silicon โข 546 GB/s | 112 GB | 24.1 GB | +87.9 GB | ~159 tok/s | โ FITS | Test Fit โ |
| Apple Mac Studio M5 Ultra (256GB Unified Memory) Apple Silicon โข 890 GB/s | 236 GB | 24.1 GB | +211.9 GB | ~259 tok/s | โ FITS | Test Fit โ |
| AMD Strix Halo Mini-PC (Ryzen AI Max+ 395 128GB) Unified APU / Mini-PC โข 256 GB/s | 110 GB | 24.1 GB | +85.9 GB | ~65 tok/s | โ FITS | Test Fit โ |
| NVIDIA DGX Spark (128GB Grace-Blackwell Unified) AI Appliance โข 273 GB/s | 118 GB | 24.1 GB | +93.9 GB | ~83 tok/s | โ FITS | Test Fit โ |
| NVIDIA H100 SXM (80GB HBM3 Cloud GPU) Datacenter Cloud GPU โข 3350 GB/s | 80 GB | 24.1 GB | +55.9 GB | ~1095 tok/s | โ FITS | Test Fit โ |