KRAFTON models
2 KRAFTON models from KRAFTON on Hugging Face, from 3.5B to 21.2B parameters, published by KRAFTON in BF16, with a 256K-token context. The smallest official model, Raon-1-nano, needs about 7.9 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 KRAFTON 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 |
|---|---|---|---|---|---|---|---|
| Raon-1-nano | 3.5B | 7.9 GB | 3.9 GB | 2.0 GB | RTX 3070 | $0.088/hr | 0.84 GB |
| A.X-K2-Raon-Speech-21B-A3B | 21.2B | 47.4 GB | 23.7 GB | 11.8 GB | RTX A6000 | $0.363/hr | not published for this architecture |
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 (2)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Raon-1-nano | 3.5B | BF16 | 7.9 GB | RTX 3070 | $0.088/hr |
| A.X-K2-Raon-Speech-21B-A3B | 21.2B | BF16 | 47.4 GB | RTX A6000 | $0.363/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.