LLM 0.32: Reasoning Traces and Server-Side Tools
LLM 0.32 ships reasoning traces to stderr, Git-style logs, server-side tools, and GPT-5.6 Luna as the new default. Upgrade guide included.
Microsoft Agent Framework Harness Is GA — The Runtime, Not the Library
Microsoft’s Agent Harness and Foundry Hosted Agents hit GA. Here’s what changed from April’s SDK ...
LinkedIn’s AI Slop Button: What Developers Must Know
LinkedIn added an "AI slop" reporting button and pulled its own AI writer. Here is ...
NVIDIA’s Open Secure AI Alliance: What Agent Devs Need
NVIDIA launched the Open Secure AI Alliance with 37+ members. Here's what's production-ready, what's still ...
Mistral Shieldstral: Open-Weights Content Moderation
Mistral's Shieldstral is a 3.8B open-weights safety classifier that beats models 7x its size. Define ...
Google’s 2025 Open Source Report: What Devs Must Know
Google published its 2025 open source retrospective this month, and most of the coverage focused on the numbers. Twenty thousand non-Google contributors. Four ...
Warp Agent CLI: Full Terminal Control for AI Agents
Warp launched a standalone Agent CLI with a pty mux architecture that lets AI agents ...
Hermes Agent v0.20: Voice, A2A Protocol, and 1.8s Startup
Nous Research shipped Hermes Agent v0.20 with streaming voice and barge-in support, A2A v1.0 multi-agent ...
MCP 2026-07-28: What Breaks and How to Migrate
MCP 2026-07-28 drops stateless architecture, deprecates sampling, roots, and logging. Here's what changed, what breaks, ...
Qwen3.8-Max Is Open Weights: Switch Your API Today
Alibaba released Qwen3.8-Max, a 2.4T MoE model beating GPT-5.6 Sol on coding. Full weights drop ...