Boole-1.7B

Boole-1.7B is a 1-bit quantized language model derived from Qwen3-1.7B that stores weights in just 247 MB, achieving 85.9% of the performance of the original BF16 model across ten benchmarks. It targets English-language inference, offering a significant reduction in model size while maintaining strong capabilities, especially in math and code tasks. It is notable for balancing efficiency and benchmark performance among small language models.