Ovi
by chetwinlow1
11.7B params · image-to-video · 298 likes · 353 downloads
Ovi is a 11.7B parameter model. At Q4 quantization it requires 6GB of VRAM. It runs comfortably on GeForce RTX 4090 (112 tok/s), GeForce RTX 5090 (168 tok/s), M4 Max 128GB (41 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| fal |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 112 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 168 tok/s | comfortable |
| M4 Max 128GB | 128GB | 41 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 20 tok/s | tight |
| M4 Pro 24GB | 24GB | 20 tok/s | tight |
| A100 PCIe 80 GB | 80GB | 206 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 414 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 99 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 82 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 79 tok/s | comfortable |