REINFORCEMENT LEARNING

FermiSense fine-tuned Qwen 3.5 9B with 2,500 GRPO steps on a single GPU for $500 and beat GPT-5.6 Sol (93%) and Opus 4.8 (91%) on automotive catalog review, reaching 97% accuracy at 68x lower cost per listing.

GRPO is suddenly the standard RL recipe for reasoning models. A no-prior-knowledge mental model of PPO, GRPO, and how DeepSeek R1's training works under the hood.