Qwen/Qwen3.5-2B
Qwen/Qwen3.5-2B
Parameters
2B
Context
256K
Downloads
4.9M
Likes
419
Architecture
qwen3_5
24 layers · 2 KV heads · head 256 · hidden 2048
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 9.7 GB RAM 32.0 · Disk 44.8 | 25.6 GB RAM 38.4 · Disk 70.8 | 122.7 GB RAM 184.1 · Disk 109.8 |
| FP88bit | 7.3 GB RAM 32.0 · Disk 41.9 | 23.2 GB RAM 34.8 · Disk 67.9 | 120.2 GB RAM 180.3 · Disk 106.9 |
| INT88bit | 7.3 GB RAM 32.0 · Disk 41.9 | 23.2 GB RAM 34.8 · Disk 67.9 | 120.2 GB RAM 180.3 · Disk 106.9 |
| AWQ4bit | 6.2 GB RAM 32.0 · Disk 40.5 | 22.0 GB RAM 33.0 · Disk 66.5 | 118.9 GB RAM 178.4 · Disk 105.5 |
| GPTQ4bit | 6.2 GB RAM 32.0 · Disk 40.5 | 22.0 GB RAM 33.0 · Disk 66.5 | 119.0 GB RAM 178.4 · Disk 105.5 |
| GGUF4bit | 6.2 GB RAM 32.0 · Disk 40.6 | 22.0 GB RAM 33.0 · Disk 66.6 | 119.0 GB RAM 178.5 · Disk 105.6 |
GPU Recommendations
1× RTX 5090
32 GB VRAM
≈ 1847.4 tok/s
Effective production throughput ≈ 1293.2 tok/s
score 0.6789
1× RTX 3090
24 GB VRAM
≈ 964.9 tok/s
Effective production throughput ≈ 675.4 tok/s
score 0.6506
8× H200 SXM
1128 GB VRAM
≈ 15835.1 tok/s
Effective production throughput ≈ 11084.6 tok/s
score 0.2197
1× RTX 4090
24 GB VRAM
≈ 1039.2 tok/s
Effective production throughput ≈ 727.4 tok/s
score 0.649
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
RTX 3090
24 GB VRAM
$89.2
/mo · vast
A10
24 GB VRAM
$171
/mo · vast
L4
24 GB VRAM
$185
/mo · vast
RTX 4090
24 GB VRAM
$190
/mo · tensordock
RTX A6000
48 GB VRAM
$264
/mo · vast