Testing AI models for parsing data (Claude vs. Gemini vs. OpenAI) | Dark Hacker News