NVIDIA Nemotron TwoTower: 2.42x Faster LLM Inference
NVIDIA’s Nemotron-Labs-TwoTower delivers 2.42x faster LLM inference at 98.7% quality without retraining the base model. Here’s what it means for your inference stack.
Microsoft Frontier Company: $2.5B to Fix What Enterprise AI Broke
Microsoft's new $2.5B Frontier Company embeds 6,000 engineers inside enterprises to fix the AI deployment ...
Moondream 3.1 on Cloudflare Workers AI: Vision Inference Under 500ms
Cloudflare added Moondream 3.1 to Workers AI — a 9B MoE vision model that runs ...
Microsoft RAMPART and Clarity: Test AI Agents in CI
Microsoft open-sourced RAMPART and Clarity to help developers test AI agents for prompt injection and ...
Bonsai 27B Runs on iPhone: The On-Device AI Tradeoffs
PrismML shrinks a 27B model to 3.9GB for iPhone. Benchmarks look solid, but developer reports ...
Google Cloud Run Sandboxes: Run Untrusted AI Code Safely
Google’s Cloud Run sandboxes are in public preview. One deployment flag gives your AI agents isolated code execution — at no extra cost ...
SK Hynix Nasdaq Debut: HBM Memory Crisis Hits Developers
SK Hynix raised $26.5B on Nasdaq July 10. Its CEO warned 2027 will be the ...
Leanstral 1.5: Mistral’s AI Found Five Real Bugs
Mistral’s Leanstral 1.5 formally verified 57 open-source repos and found 5 previously unknown bugs. Free ...
CodeQL 2.26: GitHub Scans AI Code for Prompt Injection
CodeQL 2.26.0 adds js/system-prompt-injection to detect untrusted input in AI SDK system prompts for OpenAI, ...
ZCode Is Not an IDE — Z.ai’s ADE Changes Coding
Z.ai launched ZCode, an agent-first ADE powered by GLM-5.2. It beats GPT-5.5 on SWE-bench Pro ...