🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA L40S 48GB run OmniFusion-7B (InternViT-6B-448px V1-2)?
Yes — here's how fast and up to what context, with real measured numbers.
OmniFusion-7B (InternViT-6B-448px V1-2) on the NVIDIA L40S 48GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 107k | 25.1-30 t/s | ~1.2-1.4 (slow) |
| Q8 | yes | 184k | 50.3-59.9 t/s | ~2.3-2.8 (slow) |
| Q4 | yes | 218k | 89.8-107 t/s | ~4.2-5 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
→ All models on the NVIDIA L40S 48GB
→ OmniFusion-7B (InternViT-6B-448px V1-2) specs
→ Other hardware
FAQ
Can the NVIDIA L40S 48GB run OmniFusion-7B (InternViT-6B-448px V1-2)?▶
Yes. At Q4 it generates ~89.8-107 tok/s and fits up to 218k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~4.2-5 tok/s.
How much context fits?▶
At Q4, up to about 218k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.