InternVL 2.5 is a series of advanced multimodal large language models (MLLMs). Building on InternVL 2.0, it enhances training and testing strategies and improves data quality while maintaining its core model architecture. This model integrates the newly pre-trained InternViT with various pre-trained large language models (LLMs) such as InternLM 2.5 and Qwen 2.5, using a randomly initialized MLP projector. InternVL 2.5 supports multiple images and video data, employing a dynamic high-resolution training method to enhance its capability to handle multimodal data.