Readable Rewrite — Improved hardware detection module for canirun-ai
Julien Chaumond (Hugging Face) has completed a rewrite of the hardware detection module for canirun-ai, focusing on readability and JSDoc documentation.
Sources x.com
Thomas Wolf: AI is bringing monolithic software architecture back
The Hugging Face co-founder argues that AI makes understanding and writing code cheaper, leading to a shrinking software supply chain and a revival of monolithic systems.
Sources x.com
Microsoft launches Lens — A 3.8-billion-parameter text-to-image model
Microsoft has released its Lens model on Hugging Face, a text-to-image model featuring 3.8 billion parameters that supports resolutions of up to 1440x1440.
Sources x.com
Hugging Face Storage Buckets — A cost-effective large-scale data storage solution 📦
Hugging Face Storage Buckets simplify data management when working across multiple compute providers like Azure, AWS, or Modal. This solution helps avoid expensive egress fees from traditional storage services.
Sources x.com
Hugging Face CEO: Top AI Companies Are Shifting to Training Their Own Models
Observing Cursor's aggressive push to train its own models, Hugging Face's CEO notes that serious AI companies will soon abandon API reliance to return to open source.
Sources x.com
Hugging Face Partners with Dell to Boost On-Premise AI, Relieving GPU Shortage
Hugging Face's CEO believes that open-source AI running on local/on-premise infrastructure will be the solution to GPU shortages and expensive API costs.
Sources x.com
Hugging Face Launches Ettin Reranker — A Suite of 6 Fast and Accurate Ranking Models
Hugging Face has announced Ettin Reranker, a family of 6 CrossEncoder models based on ModernBERT that improves search (RAG) accuracy while running up to 8 times faster.
Sources huggingface.co
Hugging Face's Guide to OpenAI's Privacy Filter
Hugging Face shares a method for building large-scale web applications integrated with OpenAI's Privacy Filter, helping to optimize data safety.
Sources huggingface.co
Paris 2.0 Releases Video Model Weights to the Research Community
The Paris 2.0 development team has begun selectively sharing its model weights and is looking for partners to develop video systems, world models, and robotics. The model is now available on Hugging Face.
Sources x.com
Hugging Face launches PyTorch performance optimization guide for AI engineers
Hugging Face has released part 1 of a guide on using 'torch.profiler', helping developers identify bottlenecks and reduce AI model training costs.
Sources huggingface.co
New tool simplifies sharing GPU profile traces via Hugging Face
A new command-line utility allows developers to easily share GPU profile trace files via Hugging Face, streamlining model performance analysis.
Sources x.com
hf-mem tool updates memory estimation feature for MoE models
The hf-mem tool has added a detailed breakdown of memory consumption for Mixture-of-Experts (MoE) models, helping developers optimize their infrastructure strategies.
Sources x.com
Hugging Face Champions Open Source for AI Security
Hugging Face highlights the role of transparency and open source in the future of AI security, enabling the community to detect and patch vulnerabilities faster.
Sources huggingface.co
Hugging Face CEO Urges Public Sharing of Agent Traces for Open Source AI
Hugging Face CEO Clement Delangue is advocating for the public sharing of coding and agent traces to build better open-source datasets and models.
Sources x.com
Maxime Labonne's 1B Model Trends on Hugging Face for Agentic Power
Despite having only 1B parameters, Maxime Labonne's new model is trending on Hugging Face for its surprisingly high performance on agentic tasks.
Sources x.com
Hugging Face Hub Now Renders Agent Traces Directly on Hub
Hugging Face has launched a new documentation page and rendering capability for Agent Traces on the Hub, improving transparency and debugging for AI agents.
Sources x.com
Running AI Locally: Invest in Skills Rather Than Just Staring at GPU Prices 💻
Hugging Face emphasizes that the real value of running local AI lies in the hands-on technical skills users accumulate, which far outweigh the cost of the hardware investment.
Sources x.com
Running AI Models Directly in the Browser with Transformers.js
A new guide from Hugging Face enables the integration of machine learning models that run directly inside Chrome Extensions without requiring a server.
Sources huggingface.co
IBM Announces Technical Details of Enterprise-Optimized Granite 4.1 Model Family 🚀
IBM and Hugging Face have released detailed technical documentation on the development process of the Granite 4.1 model family, a generation of LLMs specifically designed for high-performance enterprise applications.
Sources huggingface.co
Hugging Face and TII UAE Launch QIMMA — An Arabic LLM Quality Leaderboard ⛰️
In collaboration with the UAE's Technology Innovation Institute (TII), Hugging Face has introduced QIMMA, a quality-focused leaderboard aimed at standardizing the evaluation of Arabic Large Language Models.
Sources huggingface.co
vLLM Upgrades to V1: Prioritizing Accuracy to Optimize GPU Costs ⚡
ServiceNow AI and Hugging Face have officially upgraded the vLLM library from V0 to V1, focusing on improving accuracy in reinforcement learning (RL) to significantly cut infrastructure costs.
Sources huggingface.co
Hugging Face updates ASR leaderboard to prevent score gaming
Hugging Face has introduced the "Benchmaxxer Repellant" tool, which uses hidden data to prevent score gaming on its Open ASR Leaderboard.
Sources huggingface.co
Panda Master — A Robot Combining GPT and a Robotic Arm for Real-Time Calligraphy
At the Physical AI Hackathon, the 'Panda Master' project made an impression by combining the ReachyMini robot, a GPT model, and an Agilex robotic arm to converse with users and draw 'fortunes' for them.
Sources x.com
LongCat launches open-source video avatar generation model on Hugging Face
LongCat has announced a fully open-source video avatar model that enables the creation of impressive moving avatars, now available for free on Hugging Face.
Sources x.com
Hugging Face launches Reachy Mini — a powerful desktop robot for developers
Reachy Mini is a new desktop robot from Hugging Face featuring a powerful programming ecosystem, supporting agentic programming and a stable IDE environment for both professionals and children.
Sources x.com
300,000 AI experts share hardware configurations on Hugging Face
Hugging Face has released data from 300,000 users on hardware configurations for running AI, highlighting the explosive trend of local AI.
Sources x.com
LongCat launches impressive open-source talking-avatar model
LongCat has released an open-source talking-avatar model under the MIT license, achieving state-of-the-art (SOTA) performance with a live demo now available on Hugging Face.
Sources x.com
A 6-Person Team Is Developing Specialized AI Models 8 Times Faster Than OpenAI
A lean engineering team is making waves with specialized AI models that run vastly faster than those of tech giants, capturing over 500,000 downloads on Hugging Face.
Sources x.com
AWS SageMaker Integrates Strands: Deploying Open Agents from Hugging Face Hub
AWS SageMaker AI has partnered with Hugging Face to launch Strands, enabling the deployment of powerful Open Agents with MCP integration, tool use, and reasoning traces.
Sources x.com
Hugging Face launches LeRobot Humanoid — an affordable $2,500 bipedal robot
Hugging Face has introduced LeRobot Humanoid, an open-source robotics platform that enables users to build their own bipedal robots, primarily through 3D printing, at an ultra-low cost.
Sources x.com
Hugging Face Hub v1.16.0 Released: Powerful Support for Multimodal AI
The latest version of huggingface_hub officially integrates Together Compute as a new Inference provider, supporting five multimodal task types ranging from TTS to Text-to-Video.
Sources x.com
Hugging Face CEO: The future of biology requires transparent AI
Clement Delangue asserts that breakthroughs in biology and personal health should not be locked behind "black-box" paid APIs, but rather require the support of open-source AI.
Sources x.com
Hugging Face Hardware: Unveiling Real-World Data on AI Infrastructure
Hugging Face has launched its 'Hardware' page, providing real-world insights into the GPUs, CPUs, and VRAM allocations actually powering the open-source AI ecosystem.
Sources x.com
Hugging Face Updates Leaderboard to Enable Model Filtering by Parameter Count
The Hugging Face Dataset Leaderboard has added a feature to filter benchmark results by parameter range, making it easier for users to find the optimal model for their hardware capacity.
Sources x.com
Hugging Face Proposes Centralizing AI Agent 'Traces' as Memory
Caleb Fahlgren from Hugging Face highlights the importance of centralizing 'traces' (execution logs) as AI coding agents make increasingly critical decisions.
Sources x.com
Hugging Face Launches Ettin Reranker: 6 Top-tier Reranking Models
Hugging Face's new Ettin Reranker model family features 6 variants ranging from 17M to 1B parameters, setting a new benchmark for reranking performance based on ModernBERT.
Sources x.com
Hugging Face Open-Sources Genomic Foundation Models
Hugging Face has officially open-sourced its genomic foundation models, opening up opportunities for the open-source community to apply AI to DNA analysis and biomedical research.
Sources x.com
Reachy Mini Robot Officially Completed, Ready for the Robotics Community
Reachy Mini, a new open-source robot developed by Pollen Robotics in collaboration with Hugging Face, has completed assembly and is ready for real-world deployment.
Sources x.com