Small Model Hub

3,618 efficient AI models across 9 domains · Charts · API

Domain

All (3,618)Vision (1,615)3D (743)Embedding (472)Multimodal (305)Edge/Mobile (172)Code (125)Speech (89)Video (49)NLP (48)

Max Size≤10M ≤50M ≤100M ≤500M All

Max VRAM≤512MB ≤1GB ≤2GB ≤4GB ≤8GB All

Score

Raw Context (0-100)

Percentile rank within each task — comparable across domains

Showing 9 models

#	Model	Domain	Task	Params ↑	GFLOPs	VRAM	Context	Speed	Arch	Source
1	abr-ai/niagara-9m-batch.en	Speech	Speech Recognition	9.00M	—	22 MB	—	7158 RTFx	State-space model/CTC	Open ASR
2	abr-ai/niagara-19m-batch.en	Speech	Speech Recognition	20.0M	—	48 MB	—	4324 RTFx	State-space model/CTC	Open ASR
3	abr-ai/niagara-38m-batch.en	Speech	Speech Recognition	38.0M	—	91 MB	—	4015 RTFx	—	Open ASR
4	nvidia/canary-180m-flash	Speech	Speech Recognition	180.0M	—	432 MB	—	2489 RTFx	Conformer-based/Transformer	Open ASR
5	soundsgoodai/Zipformer-cr-ctc-transducer-XL-290M	Speech	Speech Recognition	290.0M	—	696 MB	—	158 RTFx	Zipformer/Pruned Stateless Transducer with k2 modified_beam_search	Open ASR
6	soundsgoodai/Zipformer-transducer-XL-290M	Speech	Speech Recognition	290.0M	—	696 MB	—	141 RTFx	Zipformer/RNN-T	Open ASR
7	AutoArk-AI/Audio8-ASR-0.1B	Speech	Speech Recognition	324.0M	—	778 MB	—	719 RTFx	Qwen3-ASR audio encoder/8-layer Qwen-style causal LM	Open ASR
8	ibm-granite/granite-speech-5.0-470m-turboctc	Speech	Speech Recognition	470.0M	—	1.1 GB	—	12946 RTFx	Streaming-style Conformer/CTC	Open ASR
9	ibm-granite/granite-speech-5.0-470m-turboctc-nc	Speech	Speech Recognition	470.0M	—	1.1 GB	—	12762 RTFx	Streaming-style Conformer/CTC	Open ASR