{"benchmark_id":"openai-mrcr:-2-needle-128k","benchmark_name":"OpenAI-MRCR: 2 needle 128k","benchmark_description":"Multi-round Co-reference Resolution (MRCR) benchmark for evaluating an LLM's ability to distinguish between multiple needles hidden in long context. Models are given a long, multi-turn synthetic conversation and must retrieve a specific instance of a repeated request, requiring reasoning and disambiguation skills beyond simple retrieval.","max_score":1.0,"categories":["long_context","reasoning"],"modality":"text","total_models":9,"entries":[{"rank":1,"model_id":"gpt-5-2025-08-07","model_name":"GPT-5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.952,"normalized_score":0.952,"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":"2025-08-07","announcement_date":"2025-08-07","multimodal":true,"param_count":null,"is_new":false},{"rank":2,"model_id":"minimax-m1-40k","model_name":"MiniMax M1 40K","organization_name":"MiniMax","organization_id":"minimax","benchmark_score":0.761,"normalized_score":0.761,"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":"2025-06-16","announcement_date":"2025-06-16","multimodal":false,"param_count":456000000000,"is_new":false},{"rank":3,"model_id":"minimax-m1-80k","model_name":"MiniMax M1 80K","organization_name":"MiniMax","organization_id":"minimax","benchmark_score":0.734,"normalized_score":0.734,"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":"2025-06-16","announcement_date":"2025-06-16","multimodal":false,"param_count":456000000000,"is_new":false},{"rank":4,"model_id":"gpt-4.1-2025-04-14","model_name":"GPT-4.1","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.572,"normalized_score":0.572,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":2.0,"output_cost_per_million":8.0,"speed_rps":100.0,"context_window":1047576,"release_date":"2025-04-14","announcement_date":"2025-04-14","multimodal":true,"param_count":null,"is_new":false},{"rank":5,"model_id":"gpt-4.1-mini-2025-04-14","model_name":"GPT-4.1 mini","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.472,"normalized_score":0.472,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":0.4,"output_cost_per_million":1.6,"speed_rps":150.0,"context_window":1047576,"release_date":"2025-04-14","announcement_date":"2025-04-14","multimodal":true,"param_count":null,"is_new":false},{"rank":6,"model_id":"gpt-4.5","model_name":"GPT-4.5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.385,"normalized_score":0.385,"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":"2025-02-27","announcement_date":"2025-02-27","multimodal":true,"param_count":null,"is_new":false},{"rank":7,"model_id":"gpt-4.1-nano-2025-04-14","model_name":"GPT-4.1 nano","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.366,"normalized_score":0.366,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":0.1,"output_cost_per_million":0.4,"speed_rps":200.0,"context_window":1047576,"release_date":"2025-04-14","announcement_date":"2025-04-14","multimodal":true,"param_count":null,"is_new":false},{"rank":8,"model_id":"gpt-4o-2024-08-06","model_name":"GPT-4o","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.319,"normalized_score":0.319,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":2.5,"output_cost_per_million":10.0,"speed_rps":132.0,"context_window":128000,"release_date":"2024-08-06","announcement_date":"2024-08-06","multimodal":true,"param_count":null,"is_new":false},{"rank":9,"model_id":"o3-mini","model_name":"o3-mini","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.187,"normalized_score":0.187,"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":"2025-01-30","announcement_date":"2025-01-30","multimodal":false,"param_count":null,"is_new":false}]}