Dots.llm1.inst MoE Model Released

Dots-studio released dots.llm1.inst, an instruction-tuned mixture-of-experts language model with 142 billion total parameters but only 14 billion activated during inference. The model matches Qwen2.5-72B performance while requiring fewer computational resources. It supports a 32,768 token context window, handles English and Chinese, and is released under the MIT license for commercial use.