{"benchmark_id":"humaneval-mul","benchmark_name":"HumanEval-Mul","benchmark_description":"A multilingual variant of the HumanEval benchmark that measures functional correctness for synthesizing programs from docstrings, consisting of 164 original programming problems assessing language comprehension, algorithms, and simple mathematics","max_score":1.0,"categories":["reasoning"],"modality":"text","total_models":2,"entries":[{"rank":1,"model_id":"deepseek-v3","model_name":"DeepSeek-V3","organization_name":"DeepSeek","organization_id":"deepseek","benchmark_score":0.826,"normalized_score":0.826,"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":"2024-12-25","announcement_date":"2024-12-25","multimodal":false,"param_count":671000000000,"is_new":false},{"rank":2,"model_id":"deepseek-v2.5","model_name":"DeepSeek-V2.5","organization_name":"DeepSeek","organization_id":"deepseek","benchmark_score":0.738,"normalized_score":0.738,"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":"2024-05-08","announcement_date":"2024-05-08","multimodal":false,"param_count":236000000000,"is_new":false}]}