XHToken models
3 XHToken models from XHToken on Hugging Face, from 1.7B to 4.1B parameters, published by XHToken in BF16, with a 1M-token context. The smallest official model, Spark-X2.5-1.7B, needs about 3.8 GB of VRAM at its published precision; the cheapest live fit is RTX 3070 at $0.088/hr.
Pick a size
One row per official XHToken size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.
| Model | Parameters | Native VRAM | FP8 VRAM | INT4 VRAM | Live GPU fit (native) | Est. $/hr | KV cache at 32K |
|---|---|---|---|---|---|---|---|
| Spark-X2.5-1.7B | 1.7B | 3.8 GB | 1.9 GB | 1.0 GB | RTX 3070 | $0.088/hr | 0.46 GB |
| Spark-X2.6-4B | 4.1B | 9.2 GB | 4.6 GB | 2.3 GB | RTX A4000 | $0.137/hr | 1.2 GB |
VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.
Official models (3)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Spark-X2.5-1.7B | 1.7B | BF16 | 3.8 GB | RTX 3070 | $0.088/hr |
| Spark-X2.5-4B | 4.1B | BF16 | 9.2 GB | RTX A4000 | $0.137/hr |
| Spark-X2.6-4B | 4.1B | BF16 | 9.2 GB | RTX A4000 | $0.137/hr |
VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.