OpenAI's o3 performance on benchmarks suggests that test-time compute is the next best way to scale AI models, raising new questions about costs and usage
Last month, AI founders and investors told TechCrunch that we're now in the “second era of scaling laws,” noting how established methods …