Internvl3

Open MLLM excelling in vision, reasoning, and long context via multimodal pre-training.