🔌
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 DGX Spark run Flan UL2?
Yes — here's how fast and up to what context, with real measured numbers.
Flan UL2 on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 285k | 5.1-6.1 t/s | ~0.2-0.2 (slow) |
| Q8 | yes | 359k | 10.2-12.2 t/s | ~0.4-0.4 (slow) |
| Q4 | yes | 392k | 18.3-21.8 t/s | ~0.7-0.8 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA DGX Spark run Flan UL2?▶
Yes. At Q4 it generates ~18.3-21.8 tok/s and fits up to 392k context; beyond that it spills to SSD swap and slows to ~0.7-0.8 tok/s.
How much context fits?▶
At Q4, up to about 392k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.