Measured by open model pull

Open models

How often an open AI model is downloaded.

#NameTrenddownloads in 30 daysGrowth

1datalab-to/chandraimage text to text model13.1k → 343k26×

2CMSManhattan/JiRackUltra_14btext generation model37.4k → 972k26×

3jinaai/jina-embeddings-v5-omni-smallfeature extraction model37k → 620k17×

4tencent/Hy3text generation model12.2k → 192k16×

5fastino/gliner2.5-multi-v1token classification model18.2k → 282k15×

6zai-org/GLM-5.3text generation model97.7k → 1.4M14×

7nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16text generation model378k → 4.7M12×

8ibm-granite/granite-4.2-8btext generation model10.4k → 128k12×

9Infomaniak-AI/vllm-translategemma-12b-itimage text to text model15.3k → 181k12×

10jinaai/jina-embeddings-v5-text-small-text-matchingsentence similarity model12k → 117k9.7×

11zai-org/GLM-5.3-Flashimage text to text model615k → 5.1M8.2×

12Comfy-Org/TRELLIS.2comfy.org37.7k → 288k7.6×

13Comfy-Org/Pixal3Dcomfy.org86.6k → 642k7.4×

14depth-anything/DA3-GIANTdepth estimation model14.5k → 100k6.9×

15jinaai/jina-embeddings-v3-hffeature extraction model15.6k → 108k6.9×

16nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16text generation model26.9k → 184k6.8×

17trklou/audio.cpp39.4k → 265k6.7×

18vibevoice/VibeVoice-1.5B-hftext to speech model11k → 71.2k6.5×

19Qwen/Qwen3.8-Flash-Nextimage text to text model193k → 1.2M6.2×

20Octen/Octen-Embedding-8Bsentence similarity model176k → 1.1M6.1×

21axolotl-ai-co/tiny-qwen2-129maxolotl.ai12.6k → 75.1k6.0×

22CMSManhattan/JiRackUltra_7btext generation model11.1k → 64.1k5.8×

23poolside/Laguna-XS.2text generation model33.1k → 188k5.7×

24nanonets/Nanonets-OCR2-3Bimage text to text model43k → 237k5.5×

25coder3101/Qwen3-VL-32B-Instruct-heretic-v2image text to text model10.7k → 58.8k5.5×

At least as big as the median of its peers · 30 days · Downloads, not users. Models released in the last year; copies and converted versions are left out.