We're sharing/showcasing best of @huggingface models. Follow to stay in loop. Promoting Open-Source models.
- Tweets 29,673
- Following 32
- Followers 68,756
- Likes 567
Meet ppt-pythia-160m-uniform250-current_mse-seed208-stage2, a fine-tuned text generation model based on EleutherAI's pythia-160m. It's small, fast, and ready for your next NLP project. Perfect for experimentation and lightweight apps.
1
518
Meet Synaptics/gemma-3-270m-it-torq: a tiny but mighty AI model optimized for Synaptics Astra SL and SL2600 MPUs. It brings large language model power to edge devices with Torq acceleration. Perfect for on-device intelligence!
1
1
4
1,389
Meet Confucius4-T3PO: a translation model built for real-time flow. It handles Chinese to English and English to Chinese simultaneously, so you can stream translations as they happen. Perfect for live chats, subtitles, and global conversations.
1
4
1,679
Meet Limite 1B Violetto: a compact text generation model with a passion for math and reasoning. Licensed Apache 2.0 and ready for vLLM, it's a lean mean problem-solving machine. Let's dive in!
2
1
13
1,785
What if you could point at anything in an image and have AI find it instantly? Meet locate-anything.cpp-gguf: an open-vocabulary object detector that runs locally. No cloud, no API keys, just pure visual grounding power. 4.5M downloads and counting!
4
4
32
2,474
Ever wanted to transcribe Russian audio? This wav2vec2 model does it automatically. Trained on Common Voice 6.0, it handles speech recognition in Russian with ease. Perfect for your next project. #AI #SpeechRecognition
1
1
5
1,837
Meet Tema_Q-X7-Thinking-i1-GGUF: a multilingual, uncensored AI model in GGUF format. It's built for agents and instruction-following, and it's ready to run locally. With 2k+ downloads, it's gaining traction fast. Let's dive in! 🧠
4
3
21
2,174
Meet Decision-Jef-0.1: a ModernBERT-based text classifier that brings calm, calibrated decisions to your AI stack. It outputs typed decisions, not just labels. Big for routing, scoring, and guardrails. 170 downloads already.
1
1
11
1,897
Meet Qwen3.6-35B-A3B Uncensored Genesis Hermes Final GGUF: a multimodal beast that sees images, reads text, and talks back without filters. 361k downloads and 643 likes say it all. This is the uncensored AI you've been waiting for.
1
30
288
15,057
Meet Cloreva-X1-0.7B, a new AI model from the community! With 0.7B params and a custom ResoNet architecture, it's built for efficiency. Perfect for edge devices and quick experiments. #AI #MachineLearning
2
1
8
2,159
Meet Starwell: a feature-extraction model built for stellarator fusion research. It turns complex plasma physics data into rich embeddings, powering flow-matching generative models. A niche but mighty tool for scientific ML.
2
3
4
2,142
Meet ThinkingCap-Qwen3.8-27B: a 27B image-text-to-text model built for token-efficient thinking. Feed it images and text together, get smart conversational answers without burning tokens. Perfect for multimodal chat apps.
1
2
16
2,537
Meet LensVLM-9B: a new vision language model built on Qwen3.5-9B. It takes images and text together, then talks back. Long context plus visual text compression means it can read dense documents and screenshots without losing detail. Perfect for builders who need real understanding, not just captions.
1
3
20
2,223
Meet Audio8-ASR-Infinite: a speech recognition model built for real-time streaming. It transcribes audio on the fly, no waiting for the full clip. Perfect for live captions, voice notes, and always-on assistants. 347 downloads and 274 likes say the community is listening.
1
13
2,125
Meet Tev1-4B-experimental, a 4B parameter decision model built on the Qwen3.5 architecture. It handles both text and image inputs, making it a versatile tool for conversational AI and beyond. Still experimental, but already turning heads.
2
3
12
2,165
Meet Al-Khwarizmi-3B-GGUF: a 3B parameter model fine-tuned from SmolLM3 for math and conversation. It's quantized in GGUF format, making it efficient to run locally. Perfect for math tutoring, problem-solving, and multilingual chat.
1
1
7
2,147
Meet Al-Khwarizmi-3B: a 3B parameter model fine-tuned for math and conversation. Built on SmolLM3, it's multilingual and ready to tackle your toughest equations while chatting naturally. Perfect for math tutors, problem solvers, and AI enthusiasts!
1
3
6
2,125
Meet Nemotron3 Nano 4B 16K GRPO Firmbeetle S120. A tiny 4B model trained with reinforcement learning to internalize token budgets. It reasons through math problems while staying within a set token limit. Efficient reasoning is the future.
3
1
5
2,140
Ever wished you could extract rich features from text without training a model from scratch? This BERT-based feature extraction model is here to help. It turns raw text into powerful embeddings ready for your ML pipeline. 158 downloads and counting.
1
4
2,157
Meet Qwen-Image-2.1-LoRAs: a collection of LoRA adapters for Qwen-Image-2.1. It lets you fine-tune text-to-image generation with lightweight modules. Perfect for custom styles, characters, or concepts without full retraining. 2,608 downloads and 66 likes show the community is already experimenting.
2
9
174
12,414
New on the scene: Qwen-Image-2.1-viggle-turbo. It's a text-to-image model that turns your words into stunning visuals. But here's the kicker: it's distilled for speed, so you get results in just a few steps. Perfect for creators who want quality without the wait.
1
2
41
4,267
Text to image just got a whole lot faster. Qwen Image 2.1 in GGUF format is here, ready to run locally on your machine. Quantized, efficient, and built for creators who want speed without losing quality. Let's dive in.
3
1
17
2,716
A Qwen image model just dropped in GGUF format and it is fully uncensored. 350k downloads and 1.4k likes say the community is loving it. Text to image without the usual filters.
2
37
454
26,318
Meet Occamy 1.0 APEX I MiniPlus V2: a GGUF image-text-to-text model with 7.5k downloads. It mixes MoE, hybrid attention, DeltaNet, and SSM for efficient multimodal magic.
1
6
2,188
Meet Occamy 1.0 APEX I MiniPlus V2.1 Abliterated GGUF. A custom quantized text generation model that's uncensored, heretic, and ready to fly. 2,548 downloads and counting.
1
7
2,208
Meet Occamy-1.0-APEX-I-MiniPlus-V2.1-GGUF: a MoE reasoning model with custom quantization. Built for llama.cpp, this GGUF is ready to run locally. Only 2,886 downloads so far, but it's a hidden gem for efficient AI.
2
2
11
2,225
Meet Qwen3.6 35B A3B MTP APEX I MiniPlus V2.1 GGUF. A MoE reasoning model with custom quantization, built for llama.cpp. 156 downloads and counting. Here's why this matters.
1
10
2,275
Meet the uncensored beast: Qwen3.8-35B-A3B-Distill-MTP-APEX-I-MiniPlus-V2.1-Abliterated. A GGUF model built to break free from alignment. 4.6k downloads and counting. Here's why it's turning heads.
9
31
278
20,816
Meet XYZ-Aquila-mini-APEX-I-MiniPlus-V2-GGUF: a multimodal AI that sees images and text, then acts like an agent. It's GGUF quantized, MoE based, and built for search and computer use. This is next level stuff.
2
3
2,175
Meet textlm, a text generation model now available in safetensors format. Whether you're building a chatbot, a story generator, or just curious about language AI, this model is a great starting point. Let's explore what it can do!
1
2
2,213
Meet XYZ-Aquila-mini-APEX-I-MiniPlus-V2.1-GGUF: a quantized MoE model built for reasoning. With custom APEX quantization and GGUF format, it runs efficiently on llama.cpp. Perfect for local AI on modest hardware.
1
1
4
2,161
New AI model alert: Thomson-1.0-Small-APEX-I-MiniPlus-V2-GGUF. It's a multimodal image-text-to-text model in GGUF format, ready for llama.cpp. Perfect for legal auditing tasks. 2,135 downloads and counting. Let's dive in!
1
10
2,180
Meet Qwen3.8-27B-Uncensored-Genesis-MTP-GGUF. A 27B MoE multimodal model that sees images and reads text, then talks back without filters. Runs locally via GGUF. 498 downloads and counting. This is the uncensored vision model you've been waiting for.
3
14
211
11,912
Meet Thomson-1.0-Small-APEX-I-MiniPlus-V2.1-GGUF: a quantized MoE reasoning model in GGUF format. Built for llama.cpp, it brings advanced reasoning to local setups. Perfect for devs who want power without cloud costs.
3
2
6
2,209
Meet T-Search-APEX-I-MiniPlus-V2.1-GGUF: a quantized MoE model built for reasoning and retrieval. Runs locally with GGUF. Perfect for search and QA tasks.
1
3
2,139
Meet ppt-pythia-1b-appendix-frequent-permuted-seed3407-stage2: a 1B param text generator fine-tuned with SFT for creative, human-like writing. With 1,911 downloads, it's quietly gaining traction. Let's explore what it can do.
1
1
8
2,156
Meet Victoria: a text generation model that's lean and mean. It uses MoE pruning, REAP, and quantization-aware training to deliver big performance with fewer resources. Perfect for efficient AI on any setup. #AI #MachineLearning
1
1
2,183
Meet Qwopus3.8-27B-Flash-V2-i1-GGUF: a multimodal AI that sees images and chats. Built on Qwen3.5, GGUF format for easy local use. 418 downloads and climbing. Exciting for vision+text fans!
1
2
10
2,443
Meet MiMo-V2.6-Distill-Qwen-9B-INT8-W8A16-AutoRound: a 9B image-text-to-text model, quantized to INT8 for speed. Built on Qwen3.5, it handles both images and text, perfect for multimodal AI apps.
1
9
2,273
Meet dQwen3.5-0.8B-Base: a tiny but mighty diffusion language model. It blends masked diffusion with hybrid attention, built on Qwen3.5. Only 744 downloads so far, but this is one to watch.
1
1
4
2,245
New diffusion language model alert! dQwen3.5-2B-Base blends masked diffusion with hybrid attention, built on Qwen3.5. It's not your typical text generator, it's a 2B parameter beast for feature extraction and generation. The future of text gen is here.
1
3
2,173
A 4B diffusion language model that generates text by refining masked tokens, not predicting left to right. Meet dQwen3.5-4B-Base: hybrid attention plus masked diffusion in one compact package.
1
1
8
2,168
Meet dQwen3.5-9B-Base: a 9B diffusion language model that blends masked diffusion with hybrid attention. It is not your usual next-token predictor. This is text generation reimagined, and it is ready to explore.
1
12
2,264
Meet dQwen3 1.7B Base: a diffusion language model built on Qwen3. Instead of predicting left to right, it generates text by refining masked tokens in parallel. A fresh take on text generation that's worth a look.
1
2
10
2,171
Diffusion models aren't just for images anymore. DiffuRefill-1B brings masked diffusion to text generation with parallel decoding. It's a fresh take on how language models can work, and it's already got 21k+ downloads.
1
2
2,166
Meet Open JEV DeBERTa V3 Large: a text classification model that doesn't just label, it makes typed decisions. Calibrated, open, and built for real-world trust. Here's why this model is turning heads.
1
1
24
2,497
