Pulse (last 3d) · Research 39 · Models 39 · Agents 24 · Open Source 16 · Legal 16 · Industry 13 · Media 12 · Products 10 · Inference 10 · Releases 7

Trending · OpenAI 11 · Apple 6 · llama.cpp 5 · Anthropic 4 · GPT-5.6 4 · xAI 4 · ChatGPT 3 · GPT-5.6 Sol 3 · Grok 4.5 3 · Prime Intellect 3 · Biomni 2 · Claude 2

Top story

xAI Grok CLI secretly uploaded entire Git repositories to Google Cloud. The tool sent gigabytes of private code before being disabled. Source

Research

AdvancedMathBench: A Benchmark Suite for Advanced Mathematical Proof Generation and Verification. Benchmark suite for evaluating LLM math proof generation and verification. Source

Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias. Mechanistic interpretability analysis of biases in LLM-as-Judge evaluation setups. Source

Physicist Yuji Tachikawa uses Claude to solve stalled string theory problem. The model spotted a calculation error and generated SymPy code. Source

Anthropic study tracks how Claude models express values across languages. Researchers analyzed 300,000 anonymized Claude conversations to observe cultural variations. Source

Tools

Anthropic launches collaborative multiplayer editing and public sharing for Claude Artifacts. Users can now edit Artifacts in real time across Slack. Source

Prime Intellect releases verifiers v1 for agentic reinforcement learning. The infrastructure trained a 100B reasoning model in two days. Source

Show HN: Jacquard, a programming language for AI-written, human-reviewed code. A programming language designed for AI-written, human-reviewed code with effect annotations. Source

Industry

Hermes agent maker Nous Research in talks for new funding at $1.5B valuation. Nous Research is negotiating funding at a $1.5B valuation. Source

The 6 wildest claims in Apple's lawsuit against OpenAI. The Verge rounds up the six wildest claims from Apple's lawsuit against OpenAI. Source

Anthropic starts localizing Claude pricing for India, its biggest market after the US. Anthropic localizes Claude pricing for India, its second-largest market. Source

Video-generation startup PixVerse raises $439M, valuation soars past $2B. PixVerse raises $439M at a $2B+ valuation for video generation. Source

Community

A Study of Microsoft's Early 2026 Rollout of Claude Code and GitHub Copilot CLI. Academic study of Microsoft's 2026 rollout of Claude Code and GitHub Copilot CLI, analyzing adoption and impact. Source

The 'agent web' is coming, where AI agents talk directly to each other instead of scraping websites. Argument for agent-native infrastructure where AI agents communicate via APIs/MCP instead of scraping UIs. Source

FutureHouse CEO Sam Rodriques warns against naive AI scaling law extrapolations. He showed 1850s steamboat math predicted 26,000 mph speeds by 2026. Source