Nanbeige models
4 Nanbeige models from Nanbeige on Hugging Face, from 848M to 33.1B parameters, published by Nanbeige in BF16, with contexts from 256K to 512K tokens. The smallest official model, Nanbeige4.2-3B-DSpark, needs about 1.9 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 Nanbeige 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 |
|---|---|---|---|---|---|---|---|
| Nanbeige4.2-3B-DSpark | 848M | 1.9 GB | 0.9 GB | 0.5 GB | RTX 3070 | $0.088/hr | 0.63 GB |
| Nanbeige4.1-3B | 3.9B | 8.8 GB | 4.4 GB | 2.2 GB | RTX A4000 | $0.173/hr | 2.0 GB |
| Nanbeige4.2-3B | 4.2B | 9.3 GB | 4.7 GB | 2.3 GB | RTX A4000 | $0.173/hr | 2.8 GB |
| Nanbeige4.3-33B | 33.1B | 74.0 GB | 37.0 GB | 18.5 GB | A100 | $1.31/hr | 6.5 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 (4)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Nanbeige4.2-3B-DSpark | 848M | BF16 | 1.9 GB | RTX 3070 | $0.088/hr |
| Nanbeige4.1-3B | 3.9B | BF16 | 8.8 GB | RTX A4000 | $0.173/hr |
| Nanbeige4.2-3B | 4.2B | BF16 | 9.3 GB | RTX A4000 | $0.173/hr |
| Nanbeige4.3-33B | 33.1B | BF16 | 74.0 GB | A100 | $1.31/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.