SmolVLM models
4 SmolVLM models on Hugging Face, from 256M to 2.2B parameters. At the precision each one is published in, the smallest needs about 0.6 GB of VRAM (SmolVLM-256M-Instruct, cheapest live fit: RTX 5060 Ti) and the largest about 10.0 GB (SmolVLM2-2.2B-Instruct, cheapest live fit: V100). The cheapest way to run SmolVLM-256M-Instruct is $0.110/hr.
SmolVLM models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| SmolVLM-256M-Instruct | 256M | BF16 | 0.6 GB | RTX 5060 Ti | $0.110/hr |
| SmolVLM-500M-Instruct | 507M | BF16 | 1.1 GB | RTX 5060 Ti | $0.110/hr |
| SmolVLM2-500M-Video-Instruct | 507M | F32 | 2.3 GB | V100 | $0.088/hr |
| SmolVLM2-2.2B-Instruct | 2.2B | F32 | 10.0 GB | V100 | $0.088/hr |
VRAM is for the precision the model is published in, with the same overhead and an 8,192-token context assumed on every page; 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.
