A user experimented with Anthropic's Claude model, testing its 'effort' setting across low, high, and max levels. The 'low' and 'high' settings provided marketing plans for the user's app, Savyit, with 'high' offering a more nuanced response. The 'max' setting, however, did not generate a new plan but instead analyzed the user's existing project files, identifying missed details and suggesting improvements based on the project's current state. This 'max' effort level was significantly more expensive due to increased tool usage and file analysis, highlighting the trade-off between depth of analysis and cost. AI
IMPACT Demonstrates how different effort settings in LLMs can lead to distinct interaction modes, from content generation to in-depth project analysis.
RANK_REASON User experiment and analysis of an existing product feature.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →