Qwen3.8-27B on NVIDIA H100 SXM 80GB: VRAM Requirement & Speed Estimate
Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.8-27B requires 19.83 GB of VRAM (16.85 GB weights + 1.34 GB KV cache + 1.64 GB overhead), fitting within H100 80G's 78.4 GB usable memory at ~137.6 tokens/s.
Physical 80 GB · Load 25.3%
Actual speed varies ~±25% by runtime
Enable Q8/Q4 KV quantization to scale batch concurrency
OLLAMA_CONTEXT_LENGTH=16384 ollama run qwen3-8-27bH100 80G Fits Qwen3.8-27B · Cost & TCO
Approx. $28,000 (~$350/GB VRAM) · 700W TDP
📊 All Quantization Levels for Qwen3.8-27B on H100 80G
| Quantization | Total VRAM (16K) | Verdict | Est. Speed | Max Context |
|---|---|---|---|---|
| FP16 / BF16 | 59.35 GB | Smooth Fit | ~42.9 tok/s | 256K |
| FP8 / Q8_0 | 32.77 GB | Smooth Fit | ~79.9 tok/s | 256K |
| GGUF Q6_K | 25.89 GB | Smooth Fit | ~102.8 tok/s | 256K |
| GGUF Q5_K_M | 22.81 GB | Smooth Fit | ~118 tok/s | 256K |
| Q4_K_M / AWQ | 19.83 GB | Smooth Fit | ~137.6 tok/s | 256K |
| GGUF IQ4_XS | 17.7 GB | Smooth Fit | ~156.2 tok/s | 256K |
| GGUF Q3_K_M | 16.5 GB | Smooth Fit | ~169.2 tok/s | 256K |
| GGUF Q2_K | 13.8 GB | Smooth Fit | ~207.7 tok/s | 256K |
💡 Recommended Alternative GPUs & Macs for Qwen3.8-27B
Frequently Asked Questions
Can NVIDIA H100 SXM 80GB run Qwen3.8-27B locally?
Yes. At Q4_K_M 4-bit quantization with a 16K context window, Qwen3.8-27B requires 19.83 GB of VRAM (16.85 GB weights + 1.34 GB KV cache + 1.64 GB overhead), fitting within H100 80G's 78.4 GB usable memory at ~137.6 tokens/s.
What is the best quantization for Qwen3.8-27B on H100 80G?
The highest-quality quantization that fits cleanly on H100 80G is FP16 / BF16 (Lossless Full Precision), using 59.35 GB VRAM at ~42.9 tokens/s with up to 256K context.
How much VRAM does Qwen3.8-27B need in full FP16 precision?
In unquantized FP16/BF16 precision (16K context), Qwen3.8-27B requires 59.35 GB VRAM (55.6 GB for weights alone).