Samsung has introduced the Tiny Recursive Model (TRM), a new AI model with just 7 million parameters that rivals or outperforms much larger language models. TRM operates with a two-layer architecture and uses a recursive self-correction loop, improving its predictions with each cycle. This approach enables the model to handle complex logic tasks efficiently without the high memory or computational demands seen in larger models. In testing, TRM achieved 87.4 percent accuracy on Sudoku-Extreme and strong results on other benchmarks, often matching or exceeding the performance of much larger models such as DeepSeek R1, Gemini 2.5 Pro, and OpenAI’s o3-mini.
