Snapshot 17 · Tuesday, 16 June 2026
The thirty models that the open-source community downloaded most this week.
Hugging Face Hub publishes a live trending list with hundreds of new entries every day. This page is the daily slice: top thirty by trendingScore (HF's velocity metric), grouped by what the model actually does. Sibling pages: llm-rate, bench, pricing, papers.
Top of the chart · text-generation
Second · image-text-to-text
Third · image-text-to-text
pipeline image-text-to-text · 10 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | moonshotai/Kimi-K2.7-Code | transformerskimi_k25image-feature-extraction | 794 | 102.2k | 762 |
| 02 | nvidia/LocateAnything-3B | transformerslocateanythingimage-feature-extraction | 2.1k | 98.7k | 328 |
| 03 | prefeitura-rio/Rio-3.5-Open-397B | transformersqwen3_5_moeimage-text-to-text | 313 | 189.7k | 301 |
| 04 | unsloth/diffusiongemma-26B-A4B-it-GGUF | gemma4unslothgemma | 286 | 120.4k | 283 |
| 05 | HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive | uncensoredqwen3.6moe | 1.9k | 2.7M | 234 |
| 06 | Jackrong/Qwopus3.6-27B-Coder-MTP-GGUF | transformersllama.cppimage-text-to-text | 218 | 79.2k | 213 |
| 07 | unsloth/Kimi-K2.7-Code-GGUF | transformerskimi_k25image-feature-extraction | 117 | 16.8k | 114 |
| 08 | unsloth/gemma-4-12b-it-GGUF | gemma4unslothgemma | 629 | 1.0M | 94 |
| 09 | DavidAU/Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-NEO-CODE-Di-IMatrix-MAX-GGUF | unslothfine tuneheretic | 368 | 366.3k | 93 |
| 10 | unsloth/MiniMax-M3-GGUF | transformersmultimodalmoe | 91 | 18.2k | 90 |
pipeline text-generation · 8 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | CohereLabs/North-Mini-Code-1.0 | transformerscohere2_moetext-generation | 410 | 12.1k | 270 |
| 02 | zai-org/GLM-5.2 | transformersglm_moe_dsatext-generation | 202 | 0 | 201 |
| 03 | OBLITERATUS/Gemma-4-12B-OBLITERATED | transformersgemma4_unifiedimage-text-to-text | 334 | 76.0k | 180 |
| 04 | microsoft/FastContext-1.0-4B-SFT | transformersqwen3text-generation | 156 | 192 | 153 |
| 05 | nex-agi/Nex-N2-Pro | transformersqwen3_5_moeimage-text-to-text | 306 | 5.0k | 142 |
| 06 | WeiboAI/VibeThinker-3B | transformersqwen2text-generation | 125 | 0 | 125 |
| 07 | nex-agi/Nex-N2-mini | transformersqwen3_5_moeimage-text-to-text | 226 | 9.2k | 120 |
| 08 | deepseek-ai/DeepSeek-V4-Pro | transformersdeepseek_v4text-generation | 4.9k | 2.8M | 118 |
pipeline any-to-any · 2 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | google/gemma-4-12B-it | transformersgemma4_unifiedimage-text-to-text | 1.1k | 1.2M | 207 |
| 02 | unsloth/gemma-4-12B-it-qat-GGUF | transformersgemma4unsloth | 250 | 305.7k | 71 |
pipeline text-to-speech · 2 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | bosonai/higgs-audio-v3-tts-4b | transformershiggs_multimodal_qwen3text-generation | 460 | 43.4k | 176 |
| 02 | Zyphra/ZONOS2 | ZONOS2text-to-speech | 97 | 539 | 97 |
pipeline uncategorised · 2 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | RazzzHF/Realism_Engine_Ideogram_4 | 107 | 0 | 90 | |
| 02 | unsloth/North-Mini-Code-1.0-GGUF | transformersunslothchat | 77 | 26.3k | 75 |
pipeline image-to-video · 1 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | zai-org/SCAIL-2 | diffuserscharacter-animationvideo-generation | 201 | 0 | 165 |
pipeline automatic-speech-recognition · 1 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | nvidia/nemotron-3.5-asr-streaming-0.6b | nemospeech-recognitioncache-aware ASR | 449 | 5.8k | 128 |
pipeline text-to-image · 1 on the trending list
| Model | Tags | Likes | Downloads | Trending | |
|---|---|---|---|---|---|
| 01 | ideogram-ai/ideogram-4-fp8 | diffuserstext-to-imageimage-generation | 558 | 12.5k | 104 |
Each morning the fleet hits the Hugging Face Hub public API at api/models?sort=trendingScore and api/models?sort=likes7d, merges the two lists, dedupes, and commits the result. The Worker reads the latest snapshot and groups by pipeline_tag. We don't filter by license, parameter count, or modality - this is the raw "what people are actually clicking on" feed. Raw JSON: today's snapshot.