{"benchmark_id":"dsbench-hard","benchmark_name":"DSBench-Hard","benchmark_description":"DSBench-Hard is DeepSeek's internal test set of difficult coding-agent problems.","max_score":1.0,"categories":["agents","code"],"modality":"text","total_models":3,"entries":[{"rank":1,"model_id":"deepseek-v4-pro-0813","model_name":"DeepSeek-V4-Pro-0813","organization_name":"DeepSeek","organization_id":"deepseek","benchmark_score":0.672,"normalized_score":0.672,"verified":false,"self_reported":true,"provider_id":"deepseek","input_cost_per_million":0.435,"output_cost_per_million":0.87,"speed_rps":null,"context_window":1048576,"release_date":"2026-08-13","announcement_date":"2026-08-13","multimodal":false,"param_count":1600000000000,"is_new":false},{"rank":2,"model_id":"deepseek-v4-flash-vision-exp","model_name":"DeepSeek-V4-Flash-Vision-Exp","organization_name":"DeepSeek","organization_id":"deepseek","benchmark_score":0.636,"normalized_score":0.636,"verified":false,"self_reported":true,"provider_id":"deepseek","input_cost_per_million":0.22,"output_cost_per_million":0.66,"speed_rps":null,"context_window":1048576,"release_date":"2026-08-21","announcement_date":"2026-08-21","multimodal":true,"param_count":null,"is_new":false},{"rank":3,"model_id":"deepseek-v4-flash-0731","model_name":"DeepSeek-V4-Flash-0731","organization_name":"DeepSeek","organization_id":"deepseek","benchmark_score":0.596,"normalized_score":0.596,"verified":false,"self_reported":true,"provider_id":"deepinfra","input_cost_per_million":0.09,"output_cost_per_million":0.18,"speed_rps":null,"context_window":1048576,"release_date":"2026-07-31","announcement_date":"2026-07-31","multimodal":false,"param_count":304000000000,"is_new":false}]}