A quantized variant of Mistral Medium 3 published by a community member, this model trades some numerical precision for reduced memory footprint using NVFP4 quantization. It behaves like a capable mid-tier text model but may show subtle quality degradation compared to the full-precision original due to aggressive quantization. Details about specific capability trade-offs at this precision level are limited.