Qwen3-VL-30B-A3B-Instruct
by Qwen
31.1B params · image-text-to-text · 579 likes · 585.1k downloads
Qwen3-VL-30B-A3B-Instruct is a 31.1B parameter model. At Q4 quantization it requires 16GB of VRAM. It runs comfortably on GeForce RTX 4090 (42 tok/s), GeForce RTX 5090 (63 tok/s), A100 PCIe 80 GB (77 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| Novita | 102 tok/s | ||
| Featherless |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 42 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 63 tok/s | comfortable |
| M4 Max 128GB | 128GB | 15 tok/s | tight |
| M4 Pro 48GB | 48GB | 7 tok/s | tight |
| M4 Pro 24GB | 24GB | 7 tok/s | tight |
| A100 PCIe 80 GB | 80GB | 77 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 155 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 37 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 30 tok/s | comfortable |
| M4 Max 64GB | 64GB | 15 tok/s | tight |