🔌
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 H200 run StarCoder 2 15B?
Yes — here's how fast and up to what context, with real measured numbers.
StarCoder 2 15B on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 375k | 116.8-139.2 t/s | ~1-1.2 (slow) |
| Q8 | yes | 432k | 233.6-278.4 t/s | ~1.9-2.3 (slow) |
| Q4 | yes | 457k | 417.1-497.1 t/s | ~3.5-4.1 (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.
FAQ
Can the NVIDIA H200 run StarCoder 2 15B?▶
Yes. At Q4 it generates ~417.1-497.1 tok/s and fits up to 457k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.5-4.1 tok/s.
How much context fits?▶
At Q4, up to about 457k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.