Models / SAGE-MM-Qwen3-VL

SAGE-MM-Qwen3-VL by allenai

GPU memory needed (smallest version of each): SAGE-MM-Qwen3-VL 4B about 15.6 GB, SAGE-MM-Qwen3-VL 8B about 24.4 GB. The cheapest runs on RTX 4060 Ti from $0.17/h right now.

2 models (4B to 8B) · ↓ 458 downloads a month · from $0.17/h on RTX 4060 Ti 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.

SAGE-MM-Qwen3-VL 4B

Version	Runtime	Weights	GPU memory	Cheapest machine now
Originalallenai/SAGE-MM-Qwen3-VL-4B-SFT	vLLM	9.7 GB	15.6 GB	$0.17/h RTX 4060 Ti
RLallenai/SAGE-MM-Qwen3-VL-4B-SFT_RL	vLLM	9.7 GB	15.6 GB	$0.17/h RTX 4060 Ti

SAGE-MM-Qwen3-VL 8B

Version	Runtime	Weights	GPU memory	Cheapest machine now
Originalallenai/SAGE-MM-Qwen3-VL-8B-SFT	vLLM	18 GB	24.4 GB	$0.46/h RTX A6000
RLallenai/SAGE-MM-Qwen3-VL-8B-SFT_RL	vLLM	18 GB	24.4 GB	$0.46/h RTX A6000

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…