A practical guide to choosing AI models within ABP Studio's AI agent system. ABP Studio supports role-based model assignment, separating the main conversation model from supporting roles like Research, Browser, Text Processor, and Git Review. The recommendation is to use a balanced model (e.g., Claude Sonnet) for everyday Ask/Plan/Agent work, reserve stronger models for complex planning or risky refactors, and assign lightweight models to background tasks like research and error summarization. Vision-capable models should be used for browser/screenshot tasks. The core principle: match model capability to task complexity rather than defaulting to the strongest model everywhere.
Table of contents
What ABP Studio Supports TodayThe Main ModelRole-Based ModelsHow I Choose In PracticeA Simple Rule76 Impressions