GLM-4.6V-Flash
by zai-org
10.3B params · image-text-to-text · 592 likes · 35.1k downloads
GLM-4.6V-Flash is a 10.3B parameter model. At Q4 quantization it requires 5GB of VRAM. It runs comfortably on GeForce RTX 4090 (127 tok/s), GeForce RTX 5090 (191 tok/s), M4 Max 128GB (46 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| Novita | 38 tok/s | ||
| Z.ai | 48 tok/s |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 127 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 191 tok/s | comfortable |
| M4 Max 128GB | 128GB | 46 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 23 tok/s | tight |
| M4 Pro 24GB | 24GB | 23 tok/s | tight |
| A100 PCIe 80 GB | 80GB | 233 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 470 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 112 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 93 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 90 tok/s | comfortable |