OSSMercury 2 First Diffusion-Based Reasoning LLM 1000+ tok/sMercury AI·high signalXBlueskyLinkedInCopy linkInception Labs Mercury 2: first diffusion-based reasoning LLM. 1000+ tok/s, 5x faster than speed-optimized LLMs. API-only but architecturally significant for agent loop economics.SourceSource pageMercury AI↳ Follow the threadPolicy dependency / Stack layerCROSS-CATEGORY: Three Independent Agent-Action Gates Shipped in 48 Hours, All Judging the Command Against Stated IntentProduct Hunt, github.com/AGGIB/Stroq and rewarelabs.com (three independent sources; the 72% figure is Reware's own)Stack layer / Threat pattern16% of 3,171 public agent-harness setups carry a confirmed security defect, and 3.8% ship a skill that pre-approves your shellarXivStack layer / Threat patternRAGFlow 0.27.2 rewrites its Agentic RAG retrieval framework and patches a starlette CVEGitHubStack layer / Threat patternqwen-code 0.23.2 turns the CLI into a remotely accessible shell with one command and a QR pairing codeGitHubStack layer / ContrastCROSS-CATEGORY: Four Independent Moves in 48 Hours All Attack the Same Line Item, the Per-Token Inference BillInception Labs, DeepSeek, Desert Ant Labs and argonautlabsai/deltafin (four independent primary sources)Stack layer / ContrastDwarkesh Patel and Jerry Han: data improvements delivered 12.0x compute efficiency to model architecture's 3.7x over six yearsDwarkesh PodcastStack layer / ContrastDesert Ant Labs Launches 18 On-Device Models Free to 100,000 Monthly Devices, With No Login and No Per-Token PriceDesert Ant Labs (single-source primary announcement; benchmark claims are the vendor's own)Stack layer / ContrastPydantic AI 2.42.0 adds a first-class provider for GitHub Copilot's OpenAI-compatible APIGitHub