MEETING_SUMMARY
by knkarthick
406M params · summarization · 196 likes · 43.4k downloads
MEETING_SUMMARY is a 406M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (3231 tok/s), GeForce RTX 5090 (4845 tok/s), M4 Max 128GB (1182 tok/s).
Inference providers
| Provider | $/1M in | $/1M out | Throughput |
|---|---|---|---|
| HF Inference |
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 3231 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 4845 tok/s | comfortable |
| M4 Max 128GB | 128GB | 1182 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 591 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 591 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 5920 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 11907 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 2856 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 2362 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 2293 tok/s | comfortable |