Nikonomics - The Economics of Small Business 171 - AI SERIES: 14 - How to Avoid Massive AI API Costs with Michael Greenberg
May 15, 2025
In this insightful discussion, Michael Greenberg, a serial entrepreneur and AI expert, reveals how he expertly navigates the world of AI for his diverse businesses. He shares his experiences in selecting AI models, including the powerful Gemini 2.5 Pro, and explains the crucial differences between standard and reasoning models. Michael dives into the significance of context windows in AI performance and recounts a cautionary tale about a $4,000 API bill, offering tips to avoid hefty charges. Don't miss his practical advice on optimizing AI for business tasks!
AI Snips
Chapters
Transcript
Episode notes
Use Reasoning Models for Business Logic
- Reasoning AI models like O1 Pro and Gemini 2.5 Pro are essential for business tasks requiring logic and math.
- These models spend more time thinking but provide better accuracy in complex problem-solving.
Invest in Expensive AI Models
- Pay for the most expensive AI models if you're seriously using AI for business writing and logic.
- Cheaper models like GPT 4.0 are less capable compared to reasoning models like O3 Mini High or O1 Pro.
Context Window Matters Most
- Context window size is critical because it determines how much conversation or data the AI can recall.
- Gemini has a context window up to a million tokens, greatly outpacing ChatGPT and Claude for complex tasks.