Qwen/Qwen3.6-27B-FP8
Qwen/Qwen3.6-27B-FP8
Parameters
21.8B
Context
256K
Downloads
8.7M
Likes
345
Architecture
qwen3_5
64 layers · 4 KV heads · head 256 · hidden 5120
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 60.1 GB RAM 90.2 · Disk 102.3 | 97.4 GB RAM 146.1 · Disk 128.3 | 308.0 GB RAM 462.0 · Disk 167.3 |
| FP88bit | 34.5 GB RAM 51.7 · Disk 70.7 | 70.5 GB RAM 105.8 · Disk 96.7 | 280.0 GB RAM 420.0 · Disk 135.7 |
| INT88bit | 34.5 GB RAM 51.7 · Disk 70.7 | 70.5 GB RAM 105.8 · Disk 96.7 | 280.0 GB RAM 420.0 · Disk 135.7 |
| AWQ4bit | 22.1 GB RAM 33.2 · Disk 55.4 | 57.6 GB RAM 86.4 · Disk 81.4 | 266.5 GB RAM 399.8 · Disk 120.4 |
| GPTQ4bit | 22.3 GB RAM 33.4 · Disk 55.6 | 57.8 GB RAM 86.7 · Disk 81.6 | 266.7 GB RAM 400.1 · Disk 120.6 |
| GGUF4bit | 22.6 GB RAM 33.9 · Disk 56.0 | 58.1 GB RAM 87.2 · Disk 82.0 | 267.1 GB RAM 400.6 · Disk 121.0 |
GPU Recommendations
1× H100 SXM
80 GB VRAM
≈ 318.1 tok/s
Effective production throughput ≈ 222.7 tok/s
score 0.6815
4× RTX 4070S Ti
64 GB VRAM
≈ 0 tok/s (N/A — benchmark unavailable)
Effective production throughput ≈ 0 tok/s
score 0.5558
8× H200 SXM
1128 GB VRAM
≈ 1458.7 tok/s
Effective production throughput ≈ 1021.1 tok/s
score 0.2197
1× A100 80GB
80 GB VRAM
≈ 193.6 tok/s
Effective production throughput ≈ 135.5 tok/s
score 0.6784
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
A100 80GB
80 GB VRAM
$876
/mo · vast
H100 SXM
80 GB VRAM
$1,178
/mo · vast
H200 SXM
141 GB VRAM
$2,555
/mo · lambda
B200
180 GB VRAM
$3,013
/mo · vast
B200 SXM
180 GB VRAM
—