InternVL3

Open MLLM family (1B-78B) from OpenGVLab. Excels at vision, reasoning, long context & agents via native multimodal pre-training. Outperforms base LLMs on text tasks.

InternVL3 is an open-source multimodal large language model (MLLM) family developed by OpenGVLab, with sizes ranging from 1B to 78B parameters. It specializes in vision, reasoning, and long-context tasks through native multimodal pre-training. Notably, it surpasses base LLMs in text-based tasks while excelling in agent-based applications, making it a versatile choice for advanced AI-driven solutions.