RTX 3090 meets the minimum VRAM requirement for Q4 inference of vikhyatk/moondream2. Review the quantization breakdown below to see how higher precision settings impact VRAM and throughput.
RTX 3090 can run vikhyatk/moondream2 with Q4 quantization. At approximately 153 tokens/second, you can expect Excellent speed - conversational response times under 1 second.
You have 20GB headroom, which is sufficient for system overhead and smooth operation.
| Quantization | VRAM needed | VRAM available | Estimated speed | Verdict |
|---|---|---|---|---|
| Q4 | 4GB | 24GB | 152.50 tok/s | ✅ Fits comfortably |
| Q8 | 7GB | 24GB | 108.53 tok/s | ✅ Fits comfortably |
| FP16 | 15GB | 24GB | 61.62 tok/s | ✅ Fits comfortably |