Hugging Face adds agentic resource discovery for search
Hugging Face announced a new agentic resource discovery feature that lets agents search for resources. The post frames it as a launch for making discovery more agent-driven.
Read EntryTischEins Radar
Curated updates for visual creators working with Midjourney, SREF codes, AI image generation, prompt systems and creative workflows.
Tuesday, June 16, 2026
GLM-5.2 Targets Long-Horizon Tasks
Hugging Face highlights GLM-5.2, a model from Z.ai, as being designed for long-horizon tasks. The post positions it as part of the growing open model ecosystem.
Read EntryFrom the Hugging Face Hub to robot hardware with Strands Agents and LeRobot
Hugging Face published a post about connecting models from the Hugging Face Hub to robot hardware using Strands Agents and LeRobot. The piece frames this as a path from hosted models to practical physical AI workflows.
Read EntryOpenAI and Molecule.one say an AI chemist improved a drug-making reaction
OpenAI and Molecule.one say a near-autonomous AI chemist using GPT-5.4 helped improve a challenging reaction used in medicinal chemistry. The work is presented as a step forward for AI-assisted research in drug development.
Read EntryMolmoMotion brings language-guided 3D motion forecasting
Hugging Face highlights MolmoMotion, a system for forecasting 3D motion using language guidance. The post presents it as a new approach to understanding and predicting motion from text cues.
Read EntryWednesday, June 10, 2026
Astrophysicist uses OpenAI Codex to help simulate black holes
OpenAI highlights how astrophysicist Chi-kwan Chan uses Codex to build black hole simulations for studying extreme physics and testing general relativity. The example shows Codex being used as a coding assistant in scientific research workflows.
Read EntryMonday, June 8, 2026
ComfyUI runs code reviews through multiple AI models to catch missed bugs
ComfyUI describes a GitHub Action that runs code reviews with four models from different labs, using two passes and one judge. The setup is meant to catch issues that a tired human reviewer or a single-lab model group might miss.
Read EntryHow an Agent Built a 3D Paris Gallery by Linking Two Hugging Face Spaces
A Hugging Face blog post shows an agent creating a 3D Paris gallery by chaining together two Hugging Face Spaces. The example focuses on combining tools in a multi-step workflow.
Read EntryTuesday, June 2, 2026
Ideogram 4.0 Gets Day-0 Support in ComfyUI
ComfyUI now supports Ideogram 4.0 on day one. The model is open weights and uses structured JSON-style prompts for image generation.
Read EntrySunday, May 31, 2026
How Gemini was used to produce Google I/O 2026
Google says its team used Gemini to help produce Google I/O 2026. The post explains the role AI played in the event creation process.
Read EntryComfyUI adds native 3D Gaussian Splat support with TripoSplat
ComfyUI now includes native support for creating 3D Gaussian splats with TripoSplat. The workflow turns a single image into 3D Gaussians inside ComfyUI.
Read EntryComfyUI May Wrapped: Integrated Updates
ComfyUI published a May wrap-up covering the integrations it added during the month. The post summarizes everything the team integrated in that period.
Read Entry