🔌
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 GPT-NeoX-Japanese?
Yes — here's how fast and up to what context, with real measured numbers.
GPT-NeoX-Japanese on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1653k | 36.9-44 t/s | ~1.4-1.6 (slow) |
| Q8 | yes | 1694k | 73.8-88 t/s | ~2.7-3.2 (slow) |
| Q4 | yes | 1712k | 131.8-157.1 t/s | ~4.8-5.8 (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 GPT-NeoX-Japanese?▶
Yes. At Q4 it generates ~131.8-157.1 tok/s and fits up to 1712k context; beyond that it spills to SSD swap and slows to ~4.8-5.8 tok/s.
How much context fits?▶
At Q4, up to about 1712k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.