We're sharing/showcasing best of @huggingface models. Follow to stay in loop. Promoting Open-Source models.
Joined December 2025
- Tweets31.8K
- Following32
- Followers71.8K
- Likes628
The community loves it: 673 likes and 5k downloads show its value. It's a powerful tool for multimodal tasks, and with the Muse-Spark lineage, you know it's built for quality. Dive in and see what you can create!
With 816B parameters, it's a heavyweight in the Muse-Spark family. Built on the transformers library and safetensors, it's designed for efficient inference. The A42B variant likely indicates a specific architecture. Get ready for top-tier performance.
This model is perfect for building conversational AI that understands images and text. Imagine chatbots that can answer questions about photos, or assistive tools for the visually impaired. The possibilities are endless!
Meet Muse-Spark-Open-816B-A42B by meta-models: a 816B parameter image-text-to-text model that's sparking conversations. With 5k+ downloads and 673 likes, it's the talk of the town. Ready to explore?
With 534k+ downloads, it's a proven favorite. Expect fast, high-quality edits that keep the essence of your original image. Whether you're a hobbyist or a pro, this model delivers top-tier results every time.
Built on a single-file diffusion architecture optimized for ComfyUI. It's a base model from the FLUX.3 family, fine-tuned for image-to-image tasks. No complex setup, just drop it in and start creating.
This model is your go-to for image editing and enhancement. Restyle photos, fix details, or create variations from a single input. Perfect for designers, artists, and anyone who wants pro-level edits without the hassle.
Meet FLUX.3 Image Klein: a ComfyUI-ready image-to-image model that edits and transforms your images with stunning precision. Over 534k downloads and 1.3k likes prove the community is loving it. Ready to see what it can do?
The value is simple: a capable 9B coder that runs locally, handles agent tasks, and won't lecture you. With 807 downloads and growing, it's a practical pick for developers who want control over their stack. Try it for your next coding agent.
Under the hood: 9B parameters, Qwen3.5 architecture, SFT-tuned for code and agent use. Shipped in GGUF format for easy local inference with llama.cpp or similar runtimes. Tags include omnicoder and tesslate, pointing to a code-focused lineage.
Use it for code generation, refactoring, and agentic tasks where you need the model to follow through without refusals. The abliterated tuning makes it useful for security research, red teaming, and any workflow where standard safety filters get in the way.
Meet Huihui-OmniCoder-9B-abliterated-GGUF: a 9B coding model built on Qwen3.5, now uncensored and ready for agent workflows. If you want a local code assistant without guardrail friction, this is worth a look.
With custom quantization and the Flint tag, this model balances size and quality. It's a niche pick with 176 downloads, but for those needing a compact, capable text generator, it delivers solid performance.
Built on the NanoPlus architecture and optimized with APEX quantization, this V2.1 release focuses on efficiency. The GGUF format ensures compatibility with llama.cpp, making it easy to load and run on consumer hardware.
Use it for creative writing, chatbots, or code generation. The GGUF format means it runs smoothly on CPU or GPU with llama.cpp. Whether you're prototyping or deploying, this model handles text tasks with ease.
