PickMiMo-V2.6-Flash-RL: 159B Multimodal LLM Understanding Video and Audio Beyond Text

This is a multimodal language model that accepts and processes images, audio, and video as inputs in addition to text. It can interpret visual information and voice data simultaneously, enabling complex understanding beyond simple text generation.

With approximately 159B parameters, it supports long-context capabilities for effectively handling long contexts. It is specialized for agent tasks, making it suitable for external tool calling and complex reasoning processes.

Based on English and Chinese, it supports FP8 and 8-bit precision quantization to ensure flexibility according to deployment environments. It can be easily loaded via the Transformers library and is freely usable in the open-source ecosystem under the MIT license.