New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the NVIDIA H200 run GTE-ModernColBERT-v1?
Yes — here's how fast and up to what context, with real measured numbers.
GTE-ModernColBERT-v1 on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1952k | 11758.4-14013.4 t/s | ~98-116.8 (slow) |
| Q8 | yes | 1954k | 23516.8-28026.8 t/s | ~196-233.6 (slow) |
| Q4 | yes | 1955k | 41994.2-50047.9 t/s | ~350-417.1 (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 GTE-ModernColBERT-v1?▶
Yes. At Q4 it generates ~41994.2-50047.9 tok/s and fits up to 1955k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~350-417.1 tok/s.
How much context fits?▶
At Q4, up to about 1955k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.