Leaderboard
The most-downloaded open-weight models on Hugging Face over the last 30 days — the community's working set, straight from the Hub and refreshed every 12 hours.
| # | Model | Downloads | Likes | Task |
|---|---|---|---|---|
| 1 | sentence-transformers/all-MiniLM-L6-v2 | 185.4M | 5K | sentence similarity |
| 2 | cross-encoder/ms-marco-MiniLM-L6-v2 | 58.3M | 265 | text ranking |
| 3 | google-bert/bert-base-uncased | 45.3M | 2.7K | fill mask |
| 4 | BAAI/bge-small-en-v1.5 | 44.3M | 491 | feature extraction |
| 5 | sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 | 37.2M | 1.3K | sentence similarity |
| 6 | google/electra-base-discriminator Google DeepMind | 31.3M | 124 | — |
| 7 | sentence-transformers/all-mpnet-base-v2 | 26.6M | 1.3K | sentence similarity |
| 8 | BAAI/bge-m3 | 24M | 3.1K | sentence similarity |
| 9 | Qwen/Qwen3-0.6B Qwen · Alibaba | 21M | 1.3K | text generation |
| 10 | openai/clip-vit-base-patch32 | 17.6M | 960 | zero shot image classification |
| 11 | FacebookAI/xlm-roberta-base | 16M | 848 | fill mask |
| 12 | hexgrad/Kokoro-82M | 13.6M | 6.3K | text to speech |
| 13 | nomic-ai/nomic-embed-text-v1.5 | 13.5M | 848 | sentence similarity |
| 14 | laion/clap-htsat-fused | 13.5M | 104 | audio classification |
| 15 | colbert-ir/colbertv2.0 | 13.1M | 360 | — |
| 16 | BAAI/bge-reranker-v2-m3 | 11.8M | 1K | text classification |
| 17 | Qwen/Qwen3-4B Qwen · Alibaba | 11.6M | 637 | text generation |
| 18 | BAAI/bge-large-en-v1.5 | 11.4M | 684 | feature extraction |
| 19 | amazon/chronos-2 | 11M | 331 | time series forecasting |
| 20 | timm/mobilenetv3_small_100.lamb_in1k | 10.4M | 77 | image classification |
| 21 | openai-community/gpt2 | 10.1M | 3.3K | text generation |
| 22 | Qwen/Qwen3-8B Qwen · Alibaba | 10M | 1.1K | text generation |
| 23 | FacebookAI/roberta-base | 10M | 614 | fill mask |
| 24 | Bingsu/adetailer | 9.9M | 723 | — |
| 25 | Qwen/Qwen2.5-7B-Instruct Qwen · Alibaba | 9.7M | 1.4K | text generation |
Leaderboard data is cached from Hugging Face Hub and refreshed every 12 hours. Scores shown are for reference; see the source for full methodology.