{"benchmark_id":"truthfulqa","name":"TruthfulQA","parent_benchmark":null,"categories":["reasoning","legal","finance","general","healthcare"],"modality":"text","multilingual":false,"max_score":1.0,"language":"en","description":"TruthfulQA is a benchmark to measure whether language models are truthful in generating answers to questions. It comprises 817 questions that span 38 categories, including health, law, finance and politics. The questions are crafted such that some humans would answer falsely due to a false belief or misconception, testing models' ability to avoid generating false answers learned from human texts.","paper_link":"https://arxiv.org/abs/2109.07958","implementation_link":null,"verified":false,"created_at":"2026-05-07T16:53:26.605516+00:00","updated_at":"2026-07-05T18:28:03.970718+00:00","statistics":{"total_models":18,"average_score":0.6054055555555555,"min_score":0.503,"max_score":0.88,"score_stddev":0.09606564833929801,"verified_count":0,"self_reported_count":18},"child_benchmarks":[],"linked_dataset":null,"models":[{"rank":1,"model_id":"mai-thinking-1","model_name":"MAI-Thinking-1","organization_id":"microsoft","organization_name":"Microsoft","organization_country":"US","score":0.88,"normalized_score":0.88,"verified":false,"self_reported":true,"self_reported_source":"https://microsoft.ai/wp-content/uploads/2026/06/main_20260602_2.pdf","analysis_method":null,"verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2026-06-02","param_count":1000000000000,"is_open_source":false,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":2,"model_id":"phi-3.5-moe-instruct","model_name":"Phi-3.5-MoE-instruct","organization_id":"microsoft","organization_name":"Microsoft","organization_country":"US","score":0.775,"normalized_score":0.775,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/microsoft/Phi-3.5-MoE-instruct","analysis_method":"10-shot","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-23","param_count":60000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":3,"model_id":"granite-3.3-8b-instruct","model_name":"Granite 3.3 8B Instruct","organization_id":"ibm","organization_name":"IBM","organization_country":"US","score":0.6686,"normalized_score":0.6686,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/ibm-granite/granite-3.3-8b-instruct","analysis_method":"Score","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2025-04-16","param_count":8000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":4,"model_id":"phi-4-mini","model_name":"Phi 4 Mini","organization_id":"microsoft","organization_name":"Microsoft","organization_country":"US","score":0.664,"normalized_score":0.664,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/microsoft/Phi-4-mini-instruct","analysis_method":"MC2, 10-shot","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2025-02-01","param_count":3840000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":5,"model_id":"phi-3.5-mini-instruct","model_name":"Phi-3.5-mini-instruct","organization_id":"microsoft","organization_name":"Microsoft","organization_country":"US","score":0.64,"normalized_score":0.64,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/microsoft/Phi-3.5-mini-instruct","analysis_method":"10-shot","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-23","param_count":3800000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":6,"model_id":"hermes-3-70b","model_name":"Hermes 3 70B","organization_id":"nous-research","organization_name":"Nous Research","organization_country":"US","score":0.6329,"normalized_score":0.6329,"verified":false,"self_reported":true,"self_reported_source":"https://nousresearch.com/","analysis_method":null,"verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-15","param_count":70000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":7,"model_id":"llama-3.1-nemotron-70b-instruct","model_name":"Llama 3.1 Nemotron 70B Instruct","organization_id":"nvidia","organization_name":"NVIDIA","organization_country":"US","score":0.5863,"normalized_score":0.5863,"verified":false,"self_reported":true,"self_reported_source":"https://developer.nvidia.com/blog/advancing-the-accuracy-efficiency-frontier-with-llama-3-1-nemotron-51b/","analysis_method":"Standard evaluation","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-10-01","param_count":70000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":8,"model_id":"qwen-2.5-14b-instruct","model_name":"Qwen2.5 14B Instruct","organization_id":"qwen","organization_name":"Alibaba Cloud / Qwen Team","organization_country":"CN","score":0.584,"normalized_score":0.584,"verified":false,"self_reported":true,"self_reported_source":"https://qwenlm.github.io/blog/qwen2.5-llm/","analysis_method":"TruthfulQA benchmark evaluation","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-09-19","param_count":14700000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":9,"model_id":"jamba-1.5-large","model_name":"Jamba 1.5 Large","organization_id":"ai21","organization_name":"AI21 Labs","organization_country":"IL","score":0.583,"normalized_score":0.583,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/ai21labs/AI21-Jamba-1.5-Large","analysis_method":"Accuracy","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-22","param_count":398000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":10,"model_id":"granite-4.0-tiny-preview","model_name":"IBM Granite 4.0 Tiny Preview","organization_id":"ibm","organization_name":"IBM","organization_country":"US","score":0.581,"normalized_score":0.581,"verified":false,"self_reported":true,"self_reported_source":"https://www.ibm.com/new/announcements/ibm-granite-4-0-tiny-preview-sneak-peek","analysis_method":"Score","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2025-05-02","param_count":7000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":11,"model_id":"qwen-2.5-32b-instruct","model_name":"Qwen2.5 32B Instruct","organization_id":"qwen","organization_name":"Alibaba Cloud / Qwen Team","organization_country":"CN","score":0.578,"normalized_score":0.578,"verified":false,"self_reported":true,"self_reported_source":"https://qwenlm.github.io/blog/qwen2.5-llm/","analysis_method":"TruthfulQA benchmark evaluation","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-09-19","param_count":32500000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":12,"model_id":"command-r-plus-04-2024","model_name":"Command R+","organization_id":"cohere","organization_name":"Cohere","organization_country":"CA","score":0.563,"normalized_score":0.563,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/CohereForAI/c4ai-command-r-plus","analysis_method":"Standardized Evaluation","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-30","param_count":104000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":13,"model_id":"qwen2-72b-instruct","model_name":"Qwen2 72B Instruct","organization_id":"qwen","organization_name":"Alibaba Cloud / Qwen Team","organization_country":"CN","score":0.548,"normalized_score":0.548,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/Qwen/Qwen2-72B","analysis_method":"Accuracy","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-07-23","param_count":72000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":14,"model_id":"qwen-2.5-coder-32b-instruct","model_name":"Qwen2.5-Coder 32B Instruct","organization_id":"qwen","organization_name":"Alibaba Cloud / Qwen Team","organization_country":"CN","score":0.542,"normalized_score":0.542,"verified":false,"self_reported":true,"self_reported_source":"https://arxiv.org/abs/2409.12186","analysis_method":"accuracy","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-09-19","param_count":32000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":15,"model_id":"jamba-1.5-mini","model_name":"Jamba 1.5 Mini","organization_id":"ai21","organization_name":"AI21 Labs","organization_country":"IL","score":0.541,"normalized_score":0.541,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/ai21labs/AI21-Jamba-1.5-Mini","analysis_method":"Accuracy","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-08-22","param_count":52000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":16,"model_id":"granite-3.3-8b-base","model_name":"Granite 3.3 8B Base","organization_id":"ibm","organization_name":"IBM","organization_country":"US","score":0.5215,"normalized_score":0.5215,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/ibm-granite/granite-3.3-8b-base","analysis_method":"Score","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2025-04-16","param_count":8170000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":17,"model_id":"qwen-2.5-coder-7b-instruct","model_name":"Qwen2.5-Coder 7B Instruct","organization_id":"qwen","organization_name":"Alibaba Cloud / Qwen Team","organization_country":"CN","score":0.506,"normalized_score":0.506,"verified":false,"self_reported":true,"self_reported_source":"https://arxiv.org/abs/2409.12186","analysis_method":"accuracy","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-09-19","param_count":7000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null},{"rank":18,"model_id":"mistral-nemo-instruct-2407","model_name":"Mistral NeMo Instruct","organization_id":"mistral","organization_name":"Mistral AI","organization_country":"FR","score":0.503,"normalized_score":0.503,"verified":false,"self_reported":true,"self_reported_source":"https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407","analysis_method":"0-shot evaluation","verification_date":null,"provider_id":null,"input_cost_per_million":null,"output_cost_per_million":null,"context_window":null,"announcement_date":"2024-07-18","param_count":12000000000,"is_open_source":true,"is_new":false,"best_latency":null,"latency_provider":null,"best_throughput":null,"throughput_provider":null,"context_provider":null}]}