Hugging Face trending rankings are a leading indicator of open-weight model adoption. This page snapshots the top trending text-generation and multimodal models as of June 2026 after the June frontier-release wave.
June 2026 Top 20 Trending (Text Generation)
| Rank | Model | Vendor | License |
|---|---|---|---|
| 1 | DeepSeek V4 Flash | DeepSeek | DeepSeek License |
| 2 | Qwen 3.7 (flagship) | Alibaba | Qwen License |
| 3 | Gemma 4 (31B Dense) | Apache 2.0 | |
| 4 | Llama 4.5 Maverick | Meta | Llama Community License |
| 5 | GLM-6 (flagship MoE) | Zhipu AI | MIT-modified |
| 6 | DeepSeek V4 (smaller variants) | DeepSeek | DeepSeek License |
| 7 | Llama 4.5 Scout | Meta | Llama Community License |
| 8 | Qwen 3.7 Coder | Alibaba | Qwen License |
| 9 | Qwen 3.7 VL (vision) | Alibaba | Qwen License |
| 10 | Kimi K2.6 | Moonshot AI | MIT-modified |
| 11 | Mistral Small 3.x | Mistral AI | Apache 2.0 |
| 12 | Hunyuan Large 3 (smaller variants) | Tencent | Custom open |
| 13 | Yi-Lightning 2 | 01.AI | Apache 2.0 |
| 14 | Phi-4 series | Microsoft | MIT |
| 15 | FLUX 1.1 Pro (image) | Black Forest Labs | Custom commercial |
| 16 | Stable Diffusion 4 | Stability AI | Custom open |
| 17 | Whisper v3 Turbo (audio) | OpenAI | MIT |
| 18 | Sentence-Transformers v3 | Multiple | Apache 2.0 |
| 19 | NV-Embed v3 (NVIDIA) | NVIDIA | Custom open |
| 20 | BGE-M3 v2 (BAAI) | BAAI | MIT |
Key Takeaways
- DeepSeek V4 Flash held the top trending slot.
- Chinese open-weight models hold five of the top ten slots, the highest concentration on record.
- Gemma 4 maintains a strong number three position thanks to Apache 2.0 commercial clarity.
- Embedding models (NV-Embed, BGE-M3, Sentence-Transformers) continue to dominate the long-tail of downloads.
Methodology
Trending rankings from huggingface.co/models trending tab, sampled at the start of June 2026. Trending reflects recent download velocity not cumulative downloads. Updated monthly.
How Presenc AI Helps
Presenc AI tracks brand visibility on the open-weight models that get embedded in developer tooling and self-hosted enterprise deployments, where brand mentions propagate into derivative outputs users never directly attribute to a specific model.