This is ForestFireVLM-7B, a finetune of Qwen2.5-VL-7B-Instruct. Our demo shows how Vision-Language Models can give detailled and structured captions for forest fires from UAV perspectives.