{"slug": "ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last", "title": "🤖 AI Agents Weekly: Microsoft's Seven MAI Models, Gemma 4 12B, NVIDIA Nemotron 3 Ultra, Agents' Last Exam, Devin Desktop, and More", "summary": "Microsoft launched seven in-house MAI models, including its first reasoning model MAI-Thinking-1, aiming for self-sufficiency from OpenAI. Google released Gemma 4 12B, an open multimodal model for agentic reasoning on consumer hardware. NVIDIA opened Nemotron 3 Ultra, a 550B parameter model, while Anthropic warned about recursive self-improvement risks.", "body_md": "# 🤖 AI Agents Weekly: Microsoft's Seven MAI Models, Gemma 4 12B, NVIDIA Nemotron 3 Ultra, Agents' Last Exam, Devin Desktop, and More\n\n### Microsoft's Seven MAI Models, Gemma 4 12B, NVIDIA Nemotron 3 Ultra, Agents' Last Exam, Devin Desktop, and More\n\nIn today’s issue:\n\nMicrosoft ships seven new MAI models\n\nMAI-Thinking-1 takes on Claude Sonnet\n\nGemma 4 12B runs agents on a laptop\n\nNVIDIA opens 550B Nemotron 3 Ultra\n\nAnthropic warns of recursive self-improvement\n\nAgents’ Last Exam stumps frontier agents\n\nClaude Platform gets an ant CLI\n\nCognition launches Devin Desktop\n\nNous ships Hermes Desktop\n\nCodex builds iOS apps end-to-end\n\nChatGPT memory learns to dream\n\nMulti-agent computer use beats solo CUAs\n\nEconomy of Minds prices agent actions\n\nLEAP solves all 12 Putnam problems\n\nA harness rewrites itself for +19 SWE points\n\nAnd all the top AI dev news, papers, and tools.\n\n**Top Stories**\n\n**Microsoft Launches Seven In-House MAI Models**\n\nMicrosoft AI unveiled a family of seven models trained from scratch, led by MAI-Thinking-1, its first reasoning model, in a bid for long-term self-sufficiency from OpenAI.\n\n**MAI-Thinking-1:** A 35B reasoning model that scores 97% on AIME and 53% on SWE-Bench Pro, with early testers preferring it side-by-side over Claude Sonnet 4.6 on overall quality.**A full stack:** The launch also ships MAI-Image-2.5 and Flash, MAI-Transcribe-1.5, MAI-Voice-2 and Flash, and MAI-Code-1-Flash for code generation.**Clean training:** Every model was trained on commercially licensed data with no distillation from third-party labs, which Microsoft frames as a hedge against legal risk for enterprise customers.**Why it matters:** Suleyman positions the release as a “hill-climbing machine,” a shared training infrastructure meant to keep Microsoft on the frontier as compute scales, and a direct shot at its biggest enterprise rival.\n\nMAI-Thinking-1 ships with a detailed 109-page technical report.\n\n**Gemma 4 12B Brings Agentic Reasoning to Your Laptop**\n\nGoogle released Gemma 4 12B, a unified, encoder-free multimodal open model that brings agentic reasoning, vision, and native audio to consumer hardware under an Apache 2.0 license.\n\n**Encoder-free design:** Vision inputs pass through a single lightweight matrix multiplication and audio is projected directly into the same space as text tokens, dropping separate modality encoders.**Runs locally:** Fits in 16GB of VRAM or unified memory, small enough for a laptop, with support across LM Studio, Ollama, and Google AI Edge Gallery.**Punches up:** Reaches performance nearing Google’s larger 26B MoE model at less than half the memory footprint, and is the first mid-sized Gemma with native audio input.**Community traction:** The release topped Hacker News, with builders showing it running on a 10-year-old Xeon CPU.", "url": "https://wpnews.pro/news/ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last", "canonical_source": "https://nlp.elvissaravia.com/p/ai-agents-weekly-microsofts-seven", "published_at": "2026-06-06 15:01:28+00:00", "updated_at": "2026-06-15 21:56:32.118831+00:00", "lang": "en", "topics": ["large-language-models", "ai-agents", "ai-research", "ai-products", "ai-infrastructure"], "entities": ["Microsoft", "Google", "NVIDIA", "Anthropic", "Cognition", "Nous", "Codex", "OpenAI"], "alternates": {"html": "https://wpnews.pro/news/ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last", "markdown": "https://wpnews.pro/news/ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last.md", "text": "https://wpnews.pro/news/ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last.txt", "jsonld": "https://wpnews.pro/news/ai-agents-weekly-microsoft-s-seven-mai-models-gemma-4-12b-nvidia-nemotron-3-last.jsonld"}}