MiniGuard-v0.1 is a newly released safety classifier from PremAI that efficiently matches the benchmark accuracy of NVIDIA's 8 billion parameter Nemotron-Guard-8B, achieving 99.5% of its accuracy while being significantly smaller, faster, and cheaper to operate. With only 0.6 billion parameters, MiniGuard is 13 times smaller, 2.5 times faster, and reduces serving costs by 67%, making it particularly advantageous for applications where safety classification is critical, such as chatbots and content generation. The model was developed using techniques like targeted synthetic data, reasoning-based distillation, model soup, and FP8 quantization, which allowed it to compress the knowledge of a large model into a smaller form without sacrificing performance. On production traffic, MiniGuard retained 91.1% of Nemotron's performance and demonstrated enhanced efficiency, making it a viable drop-in replacement that offers enterprise-grade safety checks at a fraction of the infrastructure cost. The model is open-source and available under the MIT license, aimed at addressing cost and latency challenges in AI infrastructure.