RadixArk/Kimi-K3-DSpark
RadixArk/Kimi-K3-DSpark
Parameters
4.3B
Context
1024K
Downloads
5.8M
Likes
58
Architecture
qwen3
5 layers · 16 KV heads · head 64 · hidden 7168
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 12.9 GB RAM 32.0 · Disk 51.4 | 16.7 GB RAM 32.0 · Disk 77.4 | 35.8 GB RAM 53.8 · Disk 116.4 |
| FP88bit | 7.9 GB RAM 32.0 · Disk 45.2 | 11.5 GB RAM 32.0 · Disk 71.2 | 30.4 GB RAM 45.6 · Disk 110.2 |
| INT88bit | 7.9 GB RAM 32.0 · Disk 45.2 | 11.5 GB RAM 32.0 · Disk 71.2 | 30.4 GB RAM 45.6 · Disk 110.2 |
| AWQ4bit | 5.5 GB RAM 32.0 · Disk 42.2 | 9.0 GB RAM 32.0 · Disk 68.2 | 27.7 GB RAM 41.6 · Disk 107.2 |
| GPTQ4bit | 5.5 GB RAM 32.0 · Disk 42.3 | 9.0 GB RAM 32.0 · Disk 68.3 | 27.8 GB RAM 41.7 · Disk 107.3 |
| GGUF4bit | 5.6 GB RAM 32.0 · Disk 42.3 | 9.1 GB RAM 32.0 · Disk 68.3 | 27.8 GB RAM 41.8 · Disk 107.3 |
GPU Recommendations
1× RTX 3080 Ti
12 GB VRAM
≈ 442.7 tok/s
Effective production throughput ≈ 309.9 tok/s
score 0.6942
1× Tesla V100
16 GB VRAM
≈ 0 tok/s (N/A — benchmark unavailable)
Effective production throughput ≈ 0 tok/s
score 0.6388
8× H200 SXM
1128 GB VRAM
≈ 7456.3 tok/s
Effective production throughput ≈ 5219.4 tok/s
score 0.2197
1× RTX 3080
10 GB VRAM
≈ 368.9 tok/s
Effective production throughput ≈ 258.2 tok/s
score 0.6549
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
Tesla V100
16 GB VRAM
$35.7
/mo · vast
RTX 3080
10 GB VRAM
$63.8
/mo · vast
L4
24 GB VRAM
$76.2
/mo · vast
RTX 3080 Ti
12 GB VRAM
$78.8
/mo · vast
RTX 3090
24 GB VRAM
$79.6
/mo · vast