Overview Model Architecture Pre-Training SFT: Data Preparation & Quality Control Data Quality Control SFT Training Details Phase 2 SFT for the 30B Model Reinforcement Learning: A Multi-Stage, Multi-Environment Pipeline Training Methodology The Staged Curriculum Reward Signals Foundational RL: Build the Skills Agentic RL: Learning to Act (8B / 30B) Alignment: RLHF How the Three Sizes Differ Agentic AI Infrastructure for Scalable RL Results Quantization FP8 FP4 GGUF Infrastructure Hardware Software Stack Getting Started (Transformers) Installation Basic Inference (Thinking Mode) Non-Thinking Mode Low-Effort Thinking Tool Calling Basic Tool Calling Multi-Turn with Tool Response Multi-Turn Conversations History Thinking Truncation Parsing Thinking vs. Final Answer Using with Agentic Coding Harnesses OpenCode Pi OpenHands A technical walkthrough of how we built the Granite 4.2 reasoning model family.