The Archive

Hugging Face's security.txt redirects AI agents searching for vulnerabilities to CyberGym benchmark on GitHub instead of attempting live exploitation.

Study of 3,471 uncensored open-weight models on HuggingFace (Jan 2024–Mar 2026), tracking safety guardrail removals and redistribution persistence.

NVIDIA acquires Hugging Face for $13B; OpenAI publishes postmortem on HF security incident.

PipeNetwork/minimax-h3-mlx MiniMax released MiniMax-H3 two days ago - they describe it as a "a general-purpose, omni-modal generative system", which in practice means it accepts text, images, audio and video and can use them to generate up to 15 second video clips with audio included. This Python package ports it to MLX for running on Apple Silicon. I got it running on my M5 Max MacBook Pro. I cloned the repo and ran the model like this: # First download the models uvx --from huggingface_hub hf download MiniMaxAI/MiniMax-H3 \ --include 'FL2VA/*' --exclude 'FL2VA/transformer/*' uvx --from hugg...

Cybersecurity experts told TechCrunch that one of the biggest lessons to be taken from the OpenAI hack against HuggingFace has nothing to do with AI, but traditional cybersecurity defense.

The Big Pause is coming.

AutoMegaKernel (AMK) compiles a HuggingFace Llama-family model into a single persistent cooperative CUDA kernel that runs the whole forward pass in one launch, with no per-model hand-written CUDA. The contribution is the system, not raw speed. A frozen schedule-IR validator statically certifies deadlock-freedom and race-freedom via static graph checks (not a mechanized proof), so an unsafe agent-proposed schedule is rejected before launch: across 7,160 adversarial schedules (6,091 unsafe) it had zero false-accepts and accepted all 360 real lowerings. The same source retargets sm_80/sm_90/sm_1...

looks like you can run it on any potato (A1B)! [https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF](https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF) from LiquidAI: LFM2.5 is a new family of hybrid models designed for on-device deployment. It builds on the LFM2 architecture with extended pre-training and reinforcement learning. * **On-device personal assistant**: Designed to power real-life applications, chaining tool calls, and following complex instructions on all devices. * **Compressed performance**: Competitive with much larger dense and MoE models on instruction following and agen...

Hi! Andi from Hugging Face here! My team has been working over the last few months on creating a super smooth local experience for conversations with Reachy Mini, see the video! We hope people can extend this into tons of different cool use-cases. We wrote a blog explaining how to set this up, and how to modify it for tons of different use cases. Even if you don't have a Reachy Mini, you can use this as a roadmap for amazing voice agents: [https://huggingface.co/blog/local-reachy-mini-conversation](https://huggingface.co/blog/local-reachy-mini-conversation) Hope you enjoy it!

Hello everyone. The new dataset is named MONET, is Apache 2.0 and available on HF: [https://huggingface.co/datasets/jasperai/monet](https://huggingface.co/datasets/jasperai/monet) **MONET is open, Apache 2.0-licensed image–text dataset. It was built from 2.9 billion images and refined to 104.9 million high-quality samples.** We are also publishing [a paper](https://arxiv.org/abs/2605.21272) that explains how the dataset was created if you are curious and 3 compagnions projects * [A umap to visualize the distribution](https://huggingface.co/spaces/jasperai/monet-umap) * [A retreival tool ...

a feature that was requested a lot: [https://huggingface.co/models?base\_model\_relation=base](https://huggingface.co/models?base_model_relation=base)

# [](https://huggingface.co/Qwen/Qwen-Image-Bench#model-description)Model Description Q-Judger is a vision-language model fine-tuned specifically for automated evaluation of text-to-image generated images. Given a text prompt and a generated image, the model evaluates the image on fine-grained quality criteria organized in a 3-level hierarchy and outputs structured JSON scores. * **Base Model**: Qwen3.6-27B * **Task**: Image quality evaluation / judging * **Input**: Text prompt + generated image * **Output**: Structured JSON with per-dimension scores (0 = Fail, 1 = Pass, 2 = Excel, N/A) * *...

Provided in both Safetensors and GGUFs. Safetensors, llmfan46/Gemma-4-Harmonia-31B-it-uncensored-heretic: [https://huggingface.co/llmfan46/Gemma-4-Harmonia-31B-uncensored-heretic](https://huggingface.co/llmfan46/Gemma-4-Harmonia-31B-uncensored-heretic) GGUFs, llmfan46/Gemma-4-Harmonia-31B-it-uncensored-heretic-GGUF: [https://huggingface.co/llmfan46/Gemma-4-Harmonia-31B-uncensored-heretic-GGUF](https://huggingface.co/llmfan46/Gemma-4-Harmonia-31B-uncensored-heretic-GGUF) Comes with benchmark too. Find all my models here: [HuggingFace-LLMFan46](https://huggingface.co/llmfan46/models) The o...

I don't see any threads on this model. Is it because it's dense and/or without-**reasoning**? Anyone tried this for coding? >[**Capabilities**](https://huggingface.co/ibm-granite/granite-4.1-30b) Summarization Text classification Text extraction Question-answering Retrieval Augmented Generation (RAG) Code related tasks Function-calling tasks Multilingual dialog use cases Fill-In-the-Middle (FIM) code completions Some people prefer dense in this model size range(Ex: 27B over 35B-A3B). Still no feedbacks from them here. I know that some people love Granite models. Myself...

The PrismML team really cooked with these models. They're only \~3GB in size (compared to FLUX.2 Klein 4B, which is \~16GB). Apache-2.0! Official collection on HF: [https://huggingface.co/collections/prism-ml/bonsai-image](https://huggingface.co/collections/prism-ml/bonsai-image) Link to demo: [https://huggingface.co/spaces/webml-community/bonsai-image-webgpu](https://huggingface.co/spaces/webml-community/bonsai-image-webgpu)

# MOSS-TTS-v1.5 **MOSS-TTS-v1.5** is continued from [MOSS-TTS 1.0](https://huggingface.co/OpenMOSS-Team/MOSS-TTS). It preserves the main 1.0 capabilities, including zero-shot voice cloning, long-form speech generation, token-level duration control, Pinyin/IPA pronunciation control, multilingual synthesis, and code-switching. For the full 1.0 feature walkthrough, input schema, decoding hyperparameters, and evaluation tables, please refer to the [MOSS-TTS 1.0 README](https://huggingface.co/OpenMOSS-Team/MOSS-TTS). Compared with MOSS-TTS 1.0, v1.5 focuses on the following improvements: * **St...

Safetensors, llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved: [https://huggingface.co/llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved](https://huggingface.co/llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved) GGUFs, llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved-GGUF [https://huggingface.co/llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved-GGUF](https://huggingface.co/llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-Native-MTP-Preserved-GGUF) NVFP4, llmfan46/Qwen3.5-35B-A3B-uncensored-heretic-v2-N...

MiniCPM5-1B released on HuggingFace: 1B-parameter model from CPM team, likely competitive efficiency benchmark for edge deployment.

Disclaimer: I work for Numind, the company behind this open-weight model We just released a 4B model based on Qwen3.5-4B, under Apache-2.0 license. The goal is to make information extraction from complex documents more practical with an open model: PDFs, screenshots, forms, tables, receipts, invoices, multi-page documents, and other visually structured inputs. Try it, we have a huggingface space that is completely free (you don't even have to sign-up): [https://huggingface.co/spaces/numind/NuExtract3](https://huggingface.co/spaces/numind/NuExtract3) If you ever used [NuMarkdown](https://hu...

HuggingFace benchmark datasets now support filtering by model size, enabling easier comparison of sub-32B models on tasks like SWEBench.

Built this over the past few weeks as part of a multilingual research project. Figured I'd share it here. Check it out! \~9.8M web documents across 11 languages — hi, bn, ta, te, mr, gu, kn, ml, pa, ur, en. \~8.4B tokens. CC0 license. 🤗 [https://huggingface.co/datasets/AM0908/indic-hplt-v1](https://huggingface.co/datasets/AM0908/indic-hplt-v1)

Qwen3.5 0.8B sees 2.88M monthly downloads; user reports semantic understanding, JSON parsing, and latency challenges in production workflows.

Security warning: Open-OSS/privacy-filter model on HuggingFace contains obfuscated malware payload in loader.py that executes arbitrary commands.

DeepSeek releases V4 model weights on HuggingFace, expanding open-weights frontier LLM competition.

DeepSeek V4 Flash and standard V4 weights released on Hugging Face, expanding availability for local deployment.

Kimi K2.6 released on Hugging Face; availability announcement for open-weights download.