datalab-to models
3 datalab-to models from datalab-to on Hugging Face, from 686M to 9.7B parameters, published by datalab-to in BF16, with a 256K-token context. The smallest official model, surya-ocr-2, needs about 1.5 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 datalab-to 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 |
|---|---|---|---|---|---|---|---|
| surya-ocr-2 | 686M | 1.5 GB | 0.8 GB | 0.4 GB | RTX 3070 | $0.088/hr | 0.38 GB |
| chandra-ocr-2 | 5.3B | 11.8 GB | 5.9 GB | 3.0 GB | RTX 5060 Ti | $0.145/hr | 1.0 GB |
| lift | 9.7B | 21.6 GB | 10.8 GB | 5.4 GB | RTX A5000 | $0.176/hr | 1.0 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 (3)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| surya-ocr-2 | 686M | BF16 | 1.5 GB | RTX 3070 | $0.088/hr |
| chandra-ocr-2 | 5.3B | BF16 | 11.8 GB | RTX 5060 Ti | $0.145/hr |
| lift | 9.7B | BF16 | 21.6 GB | RTX A5000 | $0.176/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.