{"podcast":{"title":"Last Week in AI","slug":"last-week-in-ai","podcast_index_feed_id":396447,"rss_url":"https://rss.art19.com/last-week-in-ai","website_url":"https://art19.com/shows/last-week-in-ai","image_url":"https://content.production.cdn.art19.com/images/d8/60/88/b2/d86088b2-d713-4824-8483-a985aa7d7f32/e4063a3a93d1635f5b88961b422beb3e4fb4feab7fa085837e15faa5db2703d1830d964620373fcc524cfeee13ef3402821ce39d8fa98fd77271c57a80e7f24d.jpeg","author":"Skynet Today","episode_count":282,"summary":"Weekly summaries of the AI news that matters!","last_synced_at":null,"page_url":"https://stenobird.com/podcast/last-week-in-ai"},"episode":{"title":"#235 - Sonnet 4.6, Deep-thinking tokens, Anthropic vs Pentagon","slug":"235-sonnet-4-6-deep-thinking-tokens-anthropic-vs-pentagon","published_at":"2026-03-03T08:00:00+00:00","page_url":"https://stenobird.com/podcast/last-week-in-ai/235-sonnet-4-6-deep-thinking-tokens-anthropic-vs-pentagon","show_page_url":"https://stenobird.com/podcast/last-week-in-ai","url":"https://rss.art19.com/episodes/2ffc750f-2f06-4af6-8fa5-439366064b2f.mp3?rss_browser=BAhJIg90cmFuc2NyaWJyBjoGRVQ%3D--952c5701c84ad333c69d5faa668f8177091704f0","audio_url":"https://rss.art19.com/episodes/2ffc750f-2f06-4af6-8fa5-439366064b2f.mp3?rss_browser=BAhJIg90cmFuc2NyaWJyBjoGRVQ%3D--952c5701c84ad333c69d5faa668f8177091704f0","summary":"A deep dive into the next generation of frontier models, focusing on Anthropic's Sonnet 4.6 and Google's Gemini 3.1 Pro performance on ARC-AGI-2. The episode also explores the technical mechanics of 'deep-thinking tokens' and the geopolitical tensions surrounding AI infrastructure and defense contracts.","meta_description":"Explore the era of reasoning models: Anthropic Sonnet 4.6, Gemini 3.1 Pro, deep-thinking tokens, and the intersection of AI and global geopolitics.","key_points":["Main idea: 'Deep-thinking tokens' serve as a measurable signal for model reasoning, where high fluctuation in intermediate layers correlates with increased accuracy","Practical takeaway: The rise of multi-agent coordinators, like Perplexity's 'Computer,' marks a shift from single-model usage to agentic orchestration","Failure mode: Distillation attacks pose a significant security risk, potentially allowing adversaries to rapidly replicate frontier model capabilities","Main idea: China is bypassing GPU constraints by using advanced packaging and networking techniques to scale 7nm/5nm wafer output","Geopolitical tension: The debate intensifies over AI labs' responsibilities regarding government contracts, specifically Anthropic's relationship with the Pentagon"],"chapters":[{"start_ms":560000,"title":"Frontier Model Benchmarks","summary":"Analysis of Sonnet 4.6 and Gemini 3.1 Pro performance on the ARC-AGI-2 reasoning benchmark."},{"start_ms":1020000,"title":"The Rise of AI Agents","summary":"Discussion on xAI's Grok 4.2 beta and the emergence of multi-agent systems like Perplexity's 'Computer'."},{"start_ms":1960000,"title":"Mechanics of Deep Thinking","summary":"A technical breakdown of how token fluctuation and Jensen-Shannon divergence can signal active model reasoning."},{"start_ms":2440000,"title":"Global Compute & Infrastructure","summary":"Examining China's chip packaging strategies and the massive capital investments in specialized AI hardware."},{"start_ms":4805000,"title":"AI Security & Geopolitics","summary":"The impact of distillation attacks and the ethical dilemmas of AI labs fulfilling defense-related contracts."}],"topics":["Anthropic Sonnet","Gemini 3.1 Pro","ARC-AGI-2","Deep-thinking tokens","AI Agents","Machine Learning Interpretability","AI Geopolitics","Model Distillation"],"duration_seconds":6108,"processing_state":"processed","actions":[{"name":"request_transcript","method":"POST","url":"https://stenobird.com/v1/public/podcasts/last-week-in-ai/episodes/235-sonnet-4-6-deep-thinking-tokens-anthropic-vs-pentagon/transcription-requests","description":"Idempotently request low-priority transcript generation for this episode."},{"name":"read_markdown","method":"GET","url":"https://stenobird.com/podcast/last-week-in-ai/235-sonnet-4-6-deep-thinking-tokens-anthropic-vs-pentagon.md","description":"Read the agent-friendly Markdown representation of this episode resource."}]}}