openai/gpt-oss-20b
openai/gpt-oss-20b
Parameters
3B
Context
128K
Downloads
8.1M
Likes
4.9K
Architecture
gpt_oss
24 layers · 8 KV heads · head 64 · hidden 2880
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 54.4 GB RAM 81.7 · Disk 100.0 | 72.4 GB RAM 108.7 · Disk 126.0 | 171.6 GB RAM 257.4 · Disk 165.0 |
| FP88bit | 29.7 GB RAM 44.5 · Disk 69.5 | 46.6 GB RAM 69.9 · Disk 95.5 | 144.6 GB RAM 216.9 · Disk 134.5 |
| INT88bit | 29.7 GB RAM 44.5 · Disk 69.5 | 46.6 GB RAM 69.9 · Disk 95.5 | 144.6 GB RAM 216.9 · Disk 134.5 |
| AWQ4bit | 17.8 GB RAM 32.0 · Disk 54.8 | 34.1 GB RAM 51.2 · Disk 80.8 | 131.6 GB RAM 197.4 · Disk 119.8 |
| GPTQ4bit | 17.9 GB RAM 32.0 · Disk 55.0 | 34.3 GB RAM 51.4 · Disk 81.0 | 131.8 GB RAM 197.7 · Disk 120.0 |
| GGUF4bit | 18.3 GB RAM 32.0 · Disk 55.4 | 34.6 GB RAM 51.9 · Disk 81.4 | 132.1 GB RAM 198.2 · Disk 120.4 |
GPU Recommendations
1× H100 SXM
80 GB VRAM
≈ 330 tok/s
Effective production throughput ≈ 231 tok/s
score 0.6032
4× Tesla V100
64 GB VRAM
≈ 0 tok/s (N/A — benchmark unavailable)
Effective production throughput ≈ 0 tok/s
score 0.5404
8× H200 SXM
1128 GB VRAM
≈ 1513.3 tok/s
Effective production throughput ≈ 1059.3 tok/s
score 0.2197
1× RTX A6000
48 GB VRAM
≈ 75.7 tok/s
Effective production throughput ≈ 53 tok/s
score 0.6807
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
RTX A6000
48 GB VRAM
$210
/mo · vast
L40S
48 GB VRAM
$341
/mo · vast
RTX PRO 6000 WS
48 GB VRAM
$482
/mo · vast
RTX PRO 6000 S
48 GB VRAM
$633
/mo · vast
A100 80GB
80 GB VRAM
$876
/mo · vast