A new framework turns physics-based expert models into interpretable fit scores, beating random forests on seismic monitoring when training data is scarce.
A researcher published the full logs behind a viral claim that a frontier model can read and edit neural network weights, drawing both interest and sharp skepticism.
A new theorem proves that joint-embedding models trained separately on different modalities converge to one representation, if their training pairs show the same scene.
An open-source agent skill turns a short AI-generated clip into a playable Three.js browser game, with every asset synthesized to match the video's look.
OpenAI is now bundling monthly API credits into ChatGPT Pro and Business subscriptions, turning consumer plans into a launchpad for building apps and agents.
A classic Wide-ResNet-50-2 backbone retrained with the RACM recipe is quietly pulling 850K+ monthly downloads on Hugging Face for image classification work.
Suno's new Model Context Protocol server lets Claude, ChatGPT, Cursor, Gemini, and Claude Code generate songs inline without API keys or tab-switching.
Google open-sourced EGM, a plugin that uses deep-context agents to draft software specs and zero-context subagents to tear them apart before any code is written.
Google's new frontier model targets autonomous cybersecurity defense and multi-day agentic coding, with early access limited to vetted defenders through the Fairwind program.
Microsoft's new managed agent harness in Foundry handles orchestration, tools, context, and sandboxing so developers can skip rebuilding the execution layer.
Artificial Analysis just rolled out AA-Video-I2V v2.1, a 1080p image-to-video benchmark with 58,000+ blind human votes across 10 use cases and 10 capabilities.