Hugging Face is an AI community platform and company providing a hub for open-source machine learning models, datasets, and demo spaces. It hosts over 500,000 models and is widely used by the AI research community.
Z.ai released GLM-5.2, a 753-billion parameter open-weights large language model optimized for long-horizon coding tasks. The model features an IndexShare architecture that reduces per-token compute F…
The Agentic Resource Discovery Specification (ARD) aims to solve the growing problem of AI clients discovering external capabilities such as tools, APIs, and agents. ARD allows clients to query for ma…
GitHub launched agent finder for Copilot, enabling AI to autonomously discover and rank relevant resources like MCP servers and agents based on plain-language task descriptions. The feature implements…
Microsoft, in collaboration with Cisco, Databricks, GitHub, GoDaddy, Google, Hugging Face, Nvidia, Salesforce, ServiceNow, and Snowflake, introduced the Agentic Resource Discovery (ARD) specification,…
Lightpanda launched Lightpanda Agent and PandaScript, a native browser agent that collapses Chrome, CDP, LLM, and agent framework into a single binary. The LLM runs at buildtime to generate a determin…
Hugging Face user Vitinello reported a login error on the Hugging Face Forums on June 17, 2026, where the `huggingface_hub` token authentication failed with a 'Bad request' error and an Amazon CloudFr…
A former OpenAI researcher is building an affordable desktop robotics research setup for under €5,000, using an industrial-grade arm, two cameras, and teleoperation, to test whether a single person ca…
Microsoft released FastContext, a system that trains a dedicated explorer subagent to handle repository exploration for coding agents. By offloading read-only searches and returning compact file-line …
SignalMesh, an open-source ambient context layer for AI agent fleets, achieves 99.97% cost reduction on context reads with 1.69µs retrieval latency. The in-memory mesh allows agents to tune into signa…
AWS has released Strands Robots, an open-source SDK that integrates Hugging Face's LeRobot stack into a single agent loop, enabling users to go from a Hub dataset to a physical robot with minimal code…
A developer seeking a blueprint for building an AI Search system received guidance from RidgeRun.ai, which shared a blog post on designing a retrieval-augmented generation system. The post covers prep…
A nine-person team at Chinese social media company Sina Weibo developed VibeThinker-3B, a 3-billion-parameter language model that matches the reasoning performance of flagship AI models hundreds of ti…
Microsoft, Google, GoDaddy, Hugging Face, and other contributors have released the Agentic Resource Discovery (ARD) specification, an open standard for dynamically discovering AI agent tools and capab…
PaddlePaddle released PP-OCRv6, a new version of its optical character recognition system, on Hugging Face. The update aims to improve accuracy and efficiency in text recognition tasks.…
A developer demonstrated the differences between foundation, instruct, and chat models using Hugging Face's SmolLM2-135M family in a free Google Colab notebook. By asking the same question to three mo…
NVIDIA Research released cuTile Rust, a tile-based system for writing memory-safe, data-race-free GPU kernels in Rust. The project extends Rust's ownership model to GPU programming, achieving up to 92…
JetBrains open-sourced Mellum2, a 12B Mixture-of-Experts coding model under Apache 2.0, designed for air-gapped and compliance-locked environments where external API calls are prohibited. The model us…
Five open-source AI tools—Amuse, MiniCPM Desk Pet, DeepSeek GUI, and two others—skip complex setup by providing desktop app experiences for local image/video generation, conversational pets, and codin…
A developer released AirLLM, an open-source tool that enables running 70-billion-parameter AI models on laptops with as little as 8GB RAM, without requiring a GPU. By using memory mapping and layer sw…
A new roadmap outlines the skills needed to become an LLM engineer by 2026, focusing on adapting and serving pretrained large language models rather than training from scratch. The five-step path cove…