This week in AI features remarkable advances in generative AI alongside corporate competition, ethics, and platform monopoly debates. Key developments: OpenAI''s GPT-4 maintains top performance with proprietary training data, but Google''s Gemini, Meta''s Llama, and Alibaba''s Qwen open-source models are rapidly improving — notably reducing hallucination without cost barriers. Meta announced plans to train Llama 4 with 10x more GPUs than Llama 3.1 (~160,000 GPU cluster, world''s largest supercomputing infrastructure), expecting to reach 600,000 GPUs by year-end; Meta maintains open-source strategy with Zuckerberg publicly criticizing closed companies like Apple. TikTok reportedly paying Microsoft $20M/month for OpenAI model access (approximately 25% of Azure OpenAI Service monthly revenue) — complicated by ByteDance''s earlier unauthorized use of OpenAI technology leading to account suspension. Microsoft added OpenAI to its annual report competitor list as "AI provider and search/news advertising competitor" following OpenAI''s SearchGPT launch, threatening Bing. Anthropic launched $100M Anthology Fund to support AI startup ecosystem. Key insight: the open-source vs. proprietary AI model competition is determining not just technical capabilities but market structures, regulatory treatment, and innovation ecosystems — with open-source models increasingly competitive while proprietary models maintain advantages in the most demanding applications.