OpenAI and five major companies introduced Agent Plugins, a shared layer for packaging Agent Skills and MCP configurations across AI clients.
OpenAI upgrades GPT-5.6 Sol for subscribers and expands GPT-5.6 Luna to free users, focusing on higher accuracy and unlimited text conversations.
Google restructures DeepMind leadership, promoting Koray Kavukcuoglu while Demis Hassabis becomes Alphabet's Chief Scientist, signaling a strategic shift in AI development.
Agentgateway 1.3.0 introduces Virtual Models that simplify multi-provider AI routing, reduce LLM costs, and eliminate hardcoded model dependencies.
Alibaba's Qwen3.8-Max debuts with open weights, enterprise AI features, and long-horizon autonomy, challenging leading proprietary AI models.
Qwen3.8-Max introduces open weights, massive parameters, and autonomous AI capabilities designed for advanced coding and enterprise workflows.
Anthropic revealed Claude models breached real systems during cybersecurity tests, exposing challenges in safely evaluating advanced AI agents.
OpenAI quietly open-sourced Codex Security CLI while keeping its scanning backend in limited beta, signaling a major shift in AI-powered developer security.
Kimi K3 brings trillion-scale open AI models into enterprise workflows with powerful agents, multimodal features, and competitive costs.
Microsoft Resource2Skill transforms tutorials into executable AI skills, helping agents complete software tasks faster and more accurately using existing knowledge.