{"benchmark_id":"tau3-banking","benchmark_name":"Tau3 Banking","benchmark_description":"τ³-Bench banking domain evaluates agentic models on multi-turn, tool-using customer-support scenarios in a simulated retail banking environment.","max_score":1.0,"categories":["reasoning","agents","tool_calling"],"modality":"text","total_models":7,"entries":[{"rank":1,"model_id":"grok-4.5","model_name":"Grok 4.5","organization_name":"xAI","organization_id":"xai","benchmark_score":0.33,"normalized_score":0.33,"verified":false,"self_reported":false,"provider_id":"xai","input_cost_per_million":2.0,"output_cost_per_million":6.0,"speed_rps":80.0,"context_window":500000,"release_date":"2026-07-16","announcement_date":"2026-07-16","multimodal":true,"param_count":null,"is_new":false},{"rank":2,"model_id":"muse-glimmer-30b","model_name":"Muse Glimmer-30B","organization_name":"Meta","organization_id":"meta","benchmark_score":0.235,"normalized_score":0.235,"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":"2026-08-10","announcement_date":"2026-08-10","multimodal":true,"param_count":29600000000,"is_new":false},{"rank":3,"model_id":"solar-pro4","model_name":"Solar Pro 4","organization_name":"Upstage","organization_id":"upstage","benchmark_score":0.23,"normalized_score":0.23,"verified":false,"self_reported":true,"provider_id":"upstage","input_cost_per_million":0.3,"output_cost_per_million":1.2,"speed_rps":null,"context_window":524288,"release_date":"2026-08-06","announcement_date":"2026-08-06","multimodal":false,"param_count":null,"is_new":false},{"rank":4,"model_id":"inkling-small","model_name":"Inkling-Small","organization_name":"Thinking Machines Lab","organization_id":"thinking-machines","benchmark_score":0.155,"normalized_score":0.155,"verified":false,"self_reported":true,"provider_id":"thinking-machines","input_cost_per_million":0.3,"output_cost_per_million":1.2,"speed_rps":null,"context_window":256000,"release_date":"2026-07-30","announcement_date":"2026-07-30","multimodal":true,"param_count":276000000000,"is_new":false},{"rank":5,"model_id":"mistral-medium-3-5","model_name":"Mistral Medium 3.5","organization_name":"Mistral AI","organization_id":"mistral","benchmark_score":0.134,"normalized_score":0.134,"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":6,"model_id":"nemotron-3.5-lightning-30b-a3b","model_name":"Nemotron 3.5 Lightning (30B A3B)","organization_name":"NVIDIA","organization_id":"nvidia","benchmark_score":0.0928,"normalized_score":0.0928,"verified":false,"self_reported":true,"provider_id":"deepinfra","input_cost_per_million":0.05,"output_cost_per_million":0.2,"speed_rps":null,"context_window":262144,"release_date":"2026-08-11","announcement_date":"2026-08-11","multimodal":false,"param_count":30000000000,"is_new":false},{"rank":7,"model_id":"lfm-2.5-2.6b","model_name":"LFM2.5-2.6B","organization_name":"Liquid AI","organization_id":"liquid","benchmark_score":0.0567,"normalized_score":0.0567,"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":"2026-08-04","announcement_date":"2026-08-04","multimodal":false,"param_count":2690000000,"is_new":false}]}