Qwen3.6-35B-A3B (MoE) on NVIDIA H100 SXM 80GB: VRAM Requirement & Speed Estimate
Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.6-35B-A3B (MoE) requires 23.96 GB of VRAM (21.83 GB weights + 0.4 GB KV cache + 1.74 GB overhead), fitting within H100 80G's 78.4 GB usable memory at ~420 tokens/s.
Physical 80 GB · Load 30.6%
MoE: reads only 1.82 GB active weights/step
Enable Q8/Q4 KV quantization to scale batch concurrency
OLLAMA_CONTEXT_LENGTH=16384 ollama run qwen3-6-35b-a3bH100 80G Fits Qwen3.6-35B-A3B (MoE) · Cost & TCO
Approx. $28,000 (~$350/GB VRAM) · 700W TDP
📊 All Quantization Levels for Qwen3.6-35B-A3B (MoE) on H100 80G
| Quantization | Total VRAM (16K) | Verdict | Est. Speed | Max Context |
|---|---|---|---|---|
| FP16 / BF16 | 75.14 GB | Tight Fit (Watch Context) | ~357.9 tok/s | 171K |
| FP8 / Q8_0 | 40.72 GB | Smooth Fit | ~420 tok/s | 256K |
| GGUF Q6_K | 31.81 GB | Smooth Fit | ~420 tok/s | 256K |
| GGUF Q5_K_M | 27.82 GB | Smooth Fit | ~420 tok/s | 256K |
| Q4_K_M / AWQ | 23.96 GB | Smooth Fit | ~420 tok/s | 256K |
| GGUF IQ4_XS | 21.21 GB | Smooth Fit | ~420 tok/s | 256K |
| GGUF Q3_K_M | 19.65 GB | Smooth Fit | ~420 tok/s | 256K |
| GGUF Q2_K | 16.16 GB | Smooth Fit | ~420 tok/s | 256K |
💡 Recommended Alternative GPUs & Macs for Qwen3.6-35B-A3B (MoE)
Frequently Asked Questions
Can NVIDIA H100 SXM 80GB run Qwen3.6-35B-A3B (MoE) locally?
Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.6-35B-A3B (MoE) requires 23.96 GB of VRAM (21.83 GB weights + 0.4 GB KV cache + 1.74 GB overhead), fitting within H100 80G's 78.4 GB usable memory at ~420 tokens/s.
What is the best quantization for Qwen3.6-35B-A3B (MoE) on H100 80G?
The highest-quality quantization that fits cleanly on H100 80G is FP16 / BF16 (Lossless Full Precision), using 75.14 GB VRAM at ~357.9 tokens/s with up to 171K context.
How much VRAM does Qwen3.6-35B-A3B (MoE) need in full FP16 precision?
In unquantized FP16/BF16 precision (16K context), Qwen3.6-35B-A3B (MoE) requires 75.14 GB VRAM (72 GB for weights alone). Because it uses a Mixture-of-Experts (MoE) architecture, all 36B parameters must sit in memory, while only 3B active parameters are read per token.