{"benchmark_id":"refcocog","benchmark_name":"RefCOCOg","benchmark_description":"RefCOCOg is a referring expression comprehension benchmark that evaluates spatial grounding in images. Given a natural language expression describing an object, the model must localize the correct region, evaluated by accuracy at a 0.5 IoU threshold. It features longer, more descriptive expressions than RefCOCO and RefCOCO+.","max_score":1.0,"categories":["multimodal","grounding","vision"],"modality":"multimodal","total_models":1,"entries":[{"rank":1,"model_id":"nova-2-omni","model_name":"Nova 2 Omni","organization_name":"Amazon","organization_id":"amazon","benchmark_score":0.863,"normalized_score":0.863,"verified":false,"self_reported":true,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"speed_rps":null,"context_window":null,"release_date":"2025-12-02","announcement_date":"2025-12-02","multimodal":true,"param_count":null,"is_new":false}]}