🎙️ EP 120: Google’s AI Invents a Cancer Therapy & Anthropic Drops a Sleeper Model

This might be the most important AI + Science story all year and no one saw it coming. Google’s new 27B Gemma model just predicted a brand new cancer drug combo… and then the lab confirmed it works. Plus, Anthropic’s smallest model just got a massive upgrade and might be the best free AI agent tool around.

We’ll talk about:

• How DeepMind’s new model helped discover a cancer immunotherapy path
• Why Claude Haiku 4.5 might be more useful than bigger models like GPT-5
• A side-by-side look at Google’s Veo 3.1 vs OpenAI’s Sora
• The wild $1 trillion spend OpenAI is gearing up for

Keywords: Claude Haiku 4.5, Gemma 27B, C2S-Scale, cancer therapy AI, Veo 3.1, Sora 2, OpenAI, Claude Code, AI drug discovery, AI agents

Links: