Florence-2 models
5 Florence-2 models on Hugging Face, from 232M to 777M parameters. At the precision each one is published in, the smallest needs about 0.5 GB of VRAM (Florence-2-base, cheapest live fit: V100) and the largest about 1.7 GB (Florence-2-large, cheapest live fit: V100). The cheapest way to run Florence-2-base is $0.088/hr.
Florence-2 models
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| Florence-2-base | 232M | F16 | 0.5 GB | V100 | $0.088/hr |
| Florence-2-base-ft | 232M | F16 | 0.5 GB | V100 | $0.088/hr |
| Florence-2-VQAJP2 | 271M | BF16 | 0.6 GB | RTX 5060 Ti | $0.110/hr |
| Florence-2-large | 777M | F16 | 1.7 GB | V100 | $0.088/hr |
| Florence-2-large | 777M | F16 | 1.7 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.
