{"benchmark_id":"healthbench-hard","benchmark_name":"HealthBench Hard","benchmark_description":"A challenging variation of HealthBench that evaluates large language models' performance and safety in healthcare through 5,000 multi-turn conversations with particularly rigorous evaluation criteria validated by 262 physicians from 60 countries","max_score":1.0,"categories":["healthcare"],"modality":"text","total_models":9,"entries":[{"rank":1,"model_id":"muse-spark","model_name":"Muse Spark","organization_name":"Meta","organization_id":"meta","benchmark_score":0.428,"normalized_score":0.428,"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":"2026-04-08","announcement_date":"2026-04-08","multimodal":true,"param_count":null,"is_new":false},{"rank":2,"model_id":"gpt-5.6-sol","model_name":"GPT-5.6 Sol","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.331,"normalized_score":0.331,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":5.0,"output_cost_per_million":30.0,"speed_rps":null,"context_window":1050000,"release_date":"2026-07-09","announcement_date":"2026-07-09","multimodal":true,"param_count":null,"is_new":false},{"rank":3,"model_id":"gpt-5.6-terra","model_name":"GPT-5.6 Terra","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.327,"normalized_score":0.327,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":2.0,"output_cost_per_million":12.0,"speed_rps":null,"context_window":1050000,"release_date":"2026-07-09","announcement_date":"2026-07-09","multimodal":true,"param_count":null,"is_new":false},{"rank":4,"model_id":"gpt-5.6-luna","model_name":"GPT-5.6 Luna","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.32,"normalized_score":0.32,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":0.2,"output_cost_per_million":1.2,"speed_rps":null,"context_window":1050000,"release_date":"2026-07-09","announcement_date":"2026-07-09","multimodal":true,"param_count":null,"is_new":false},{"rank":5,"model_id":"gpt-oss-120b","model_name":"GPT OSS 120B","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.3,"normalized_score":0.3,"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":"2025-08-05","announcement_date":"2025-08-05","multimodal":false,"param_count":116800000000,"is_new":false},{"rank":6,"model_id":"gpt-5.3-chat-latest","model_name":"GPT-5.3 Chat","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.259,"normalized_score":0.259,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":1.75,"output_cost_per_million":14.0,"speed_rps":100.0,"context_window":128000,"release_date":"2026-03-04","announcement_date":"2026-03-04","multimodal":true,"param_count":null,"is_new":false},{"rank":7,"model_id":"gpt-5.5-instant","model_name":"GPT-5.5 Instant","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.229,"normalized_score":0.229,"verified":false,"self_reported":true,"provider_id":"openai","input_cost_per_million":5.0,"output_cost_per_million":30.0,"speed_rps":100.0,"context_window":400000,"release_date":"2026-05-05","announcement_date":"2026-05-05","multimodal":true,"param_count":null,"is_new":false},{"rank":8,"model_id":"gpt-oss-20b","model_name":"GPT OSS 20B","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.108,"normalized_score":0.108,"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":"2025-08-05","announcement_date":"2025-08-05","multimodal":false,"param_count":20900000000,"is_new":false},{"rank":9,"model_id":"gpt-5-2025-08-07","model_name":"GPT-5","organization_name":"OpenAI","organization_id":"openai","benchmark_score":0.016,"normalized_score":0.016,"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}]}