{"benchmark_id":"collie","benchmark_name":"COLLIE","benchmark_description":"COLLIE is a grammar-based framework for systematic construction of constrained text generation tasks. It allows specification of rich, compositional constraints across diverse generation levels and modeling challenges including language understanding, logical reasoning, and semantic planning. The COLLIE-v1 dataset contains 2,080 instances across 13 constraint structures.","max_score":1.0,"categories":["language","reasoning","writing"],"modality":"text","total_models":10,"entries":[{"rank":1,"model_id":"gpt-5-2025-08-07","model_name":"GPT-5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.99,"normalized_score":0.99,"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":"o3-mini","model_name":"o3-mini","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.987,"normalized_score":0.987,"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},{"rank":3,"model_id":"o3-2025-04-16","model_name":"o3","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.984,"normalized_score":0.984,"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":"2025-04-16","announcement_date":"2025-04-16","multimodal":true,"param_count":null,"is_new":false},{"rank":4,"model_id":"mistral-medium-3-5","model_name":"Mistral Medium 3.5","organization_name":"Mistral AI","organization_id":"mistral","benchmark_score":0.958,"normalized_score":0.958,"verified":false,"self_reported":true,"provider_id":"mistral","input_cost_per_million":1.5,"output_cost_per_million":7.5,"speed_rps":null,"context_window":256000,"release_date":"2026-04-29","announcement_date":"2026-04-29","multimodal":true,"param_count":128000000000,"is_new":false},{"rank":5,"model_id":"gpt-4.5","model_name":"GPT-4.5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.723,"normalized_score":0.723,"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":6,"model_id":"gpt-4.1-2025-04-14","model_name":"GPT-4.1","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.658,"normalized_score":0.658,"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":7,"model_id":"mistral-small-latest","model_name":"Mistral Small 4","organization_name":"Mistral AI","organization_id":"mistral","benchmark_score":0.629,"normalized_score":0.629,"verified":false,"self_reported":true,"provider_id":"mistral","input_cost_per_million":0.15,"output_cost_per_million":0.6,"speed_rps":null,"context_window":256000,"release_date":"2026-03-16","announcement_date":"2026-03-16","multimodal":true,"param_count":119000000000,"is_new":false},{"rank":8,"model_id":"gpt-4o-2024-08-06","model_name":"GPT-4o","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.61,"normalized_score":0.61,"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":"gpt-4.1-mini-2025-04-14","model_name":"GPT-4.1 mini","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.546,"normalized_score":0.546,"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":10,"model_id":"gpt-4.1-nano-2025-04-14","model_name":"GPT-4.1 nano","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.425,"normalized_score":0.425,"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}]}