Mamba models
3 Mamba models from state-spaces on Hugging Face, from 129M to 2.8B parameters, published by state-spaces in F32. The smallest official model, mamba-130m-hf, needs about 0.6 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 Mamba 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 |
|---|---|---|---|---|---|---|---|
| mamba-130m-hf | 129M | 0.6 GB | 0.1 GB | 0.1 GB | RTX 3070 | $0.088/hr | not published for this architecture |
| mamba-1.4b-hf | 1.4B | 6.1 GB | 1.5 GB | 0.8 GB | RTX 3070 | $0.088/hr | not published for this architecture |
| mamba-2.8b-hf | 2.8B | 12.4 GB | 3.1 GB | 1.5 GB | V100 | $0.110/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 (3)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| mamba-130m-hf | 129M | F32 | 0.6 GB | RTX 3070 | $0.088/hr |
| mamba-1.4b-hf | 1.4B | F32 | 6.1 GB | RTX 3070 | $0.088/hr |
| mamba-2.8b-hf | 2.8B | F32 | 12.4 GB | V100 | $0.110/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.