AIMv2-3B.

Appleopen-source2.7B paramsVision Transformer (Autoregressive Pre-trained)

Multimodal autoregressive pre-training of large vision encoder. 2.7B params, patch14, 448px resolution. Trained with image+text autoregressive objective on proprietary data. Released Nov 2024. Paper: arxiv:2411.14402.

§ 01 · Benchmarks

Every benchmark AIMv2-3B has a recorded score for.

#	Benchmark	Area · Task	Metric	Value	Rank	Date	Source
01	ImageNet-1K	Computer Vision · Image Classification	top-1-accuracy	89.5%	#4/20	—	source ↗

Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.

§ 02 · Strengths by area

Where AIMv2-3B actually performs.

Computer Vision

benchmark

avg rank #4.0

§ 05 · Sources & freshness

Where these numbers come from.

arxiv-paper

result

0 of 1 rows marked verified.