Classification models

26 classification models. Each section says what the model is, whether it may be used commercially and how fast it is; licences and checks are under Details.

Input: one RGB image. Output: ImageNet-1k class logits.

Measured on RTX 5060 Ti, torch 2.14.1+cu130; fastforks bench MODEL --size HxW measures yours. The summary uses one input, 224x224: kernels time with its speed-up over the faster original run, and kernels peak reserved memory with its ratio to the smaller original one. Where measured (one variant per family), a variant's table also lists portable under torch.compile and the fp16 TensorRT engine of portable (export; memory: the engine's device memory). – = not measured.

Other devices: Apple M1 Pro 16 GB, NVIDIA L4, NVIDIA L40S, NVIDIA A10, NVIDIA A100-SXM4-40GB, NVIDIA RTX PRO 6000; across devices.

All models

Modes, licences, training data and outputs per model

Commercial status per variant: yes if the code and the weights licence both allow commercial use (the training data is listed but does not change it; weights without a licence of their own are assumed to carry the code licence); no if either forbids it; unclear for custom licence terms (treat as no).

model	modes	code licence	weights licence	training data	commercial	outputs
alexnet	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
convnextv2_base	portable, kernels	Apache-2.0	CC-BY-NC-4.0	imagenet-21k, imagenet-1k	no (weights: CC-BY-NC-4.0)	logits
deit_base_patch16_224	portable, kernels	Apache-2.0	Apache-2.0	imagenet-1k	yes	logits
densenet121	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
dinov2_vitl14_lc	portable, kernels	Apache-2.0	Apache-2.0	lvd-142m, imagenet-1k	yes	logits
efficientnet_b0	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
efficientnet_b4	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
efficientnet_v2_s	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
efficientvit_b2	portable, kernels	Apache-2.0	Apache-2.0	imagenet-1k	yes	logits
eva02_large_patch14_448	portable, kernels	Apache-2.0	MIT	merged-38m, imagenet-21k, imagenet-1k	yes	logits
fastvit_sa24	portable, kernels	Apache-2.0	LicenseRef-Apple-Sample-Code	imagenet-1k	yes	logits
inception_v3	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
mobilenet_v2	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
mobilenet_v3_large	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
mobilenetv1_100	portable, kernels	Apache-2.0	Apache-2.0	imagenet-1k	yes	logits
mobilenetv4_conv_medium	portable, kernels	Apache-2.0	Apache-2.0	imagenet-1k	yes	logits
regnety_160	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
resnet152	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
resnet18	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
resnet50	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
shufflenet_v2_x1_0	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
squeezenet1_1	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
swin_base_patch4_window7_224	portable, kernels	Apache-2.0	MIT	imagenet-21k, imagenet-1k	yes	logits
vgg16	portable, kernels	BSD-3-Clause	BSD-3-Clause (assumed from code)	imagenet-1k	yes	logits
vit_base_patch16_224	portable, kernels	Apache-2.0	Apache-2.0	imagenet-21k, imagenet-1k	yes	logits
vit_so400m_patch14_siglip_378	portable, kernels	Apache-2.0	Apache-2.0	webli, imagenet-1k	yes	logits

Training datasets