Models / granite-4.0

granite-4.0 by ibm-granite

3 models (350M to 3B) · ↓ 141.3K downloads a month · from $0.10/h on RTX 3060 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.

granite-4.0 350M

Version	Runtime	Weights	GPU memory	Cheapest machine now
Instructibm-granite/granite-4.0-350m	vLLM	700 MB	3.3 GB	$0.10/h RTX 3060

granite-4.0 1B

Version	Runtime	Weights	GPU memory	Cheapest machine now
Originalibm-granite/granite-4.0-1b-base	vLLM	3.3 GB	6.6 GB	$0.10/h RTX 3060
Instructibm-granite/granite-4.0-1b	vLLM	3.3 GB	6.6 GB	$0.10/h RTX 3060

granite-4.0 3B

Version	Runtime	Weights	GPU memory	Cheapest machine now
Originalibm-granite/granite-4.0-3b-vision	vLLM	9.4 GB	13.5 GB	$0.16/h RTX 4060 Ti

Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").

Loading…