15 AI Updates You Missed This Week (Nemotron, Gemma 4, MAI & More)

Fifteen of the biggest AI releases this week, packed into one fast rundown. NVIDIA's 550B Nemotron 3 Ultra with a 1M-token context, Google's encoder-free Gemma 4 12B, seven new Microsoft MAI models, MiniMax M3, GPT-5.5 and Codex on Bedrock, plus new agent tooling, security harnesses, and benchmarks. https://mer.vin/2026/06/ai-engineering-roundup-june-2026-nemotron-gemma-mai-m3-bedrock-codex-and-agent-security/ Everything you need to stay current without reading 15 blog posts. ⏱️ Chapters 0:00 Intro & lineup 0:38 NVIDIA Nemotron 3 Ultra 1:00 Google Gemma 4 12B 1:22 Microsoft's 7 MAI models 1:45 MiniMax M3 2:05 GPT-5.5 + Codex on Bedrock 2:26 Codex role plugins 2:49 Hugging Face hf CLI 3:09 Holo 3.1 3:29 JetBrains Mellum2 3:48 Anthropic defending-code 4:10 Claude containment 4:31 EVA-Bench 2.0 4:52 GPT-Rosalind 5:11 MITRE AI threat mapping 5:31 Meta Muse Spark watch 👍 Like, share & subscribe for daily AI updates. #AInews #Nemotron #Gemma4 #MicrosoftMAI #MiniMax This video presents a roundup of recent "ai news", covering significant advancements in "artificial intelligence" from various developers. We discuss the latest updates to models like GPT 5.5 and Codex, and crucial "claude ai" developments. Additionally, the video explores how "ai agents" are evolving with new functionalities, impacting the broader "llm" landscape.