τ²-Bench Airline leaderboard: model accuracy, cost, time, and output tokens per task by provider
#	Model		Std dev
1	Anthropic: Claude Fable 5 Pareto	79.9%	±2.8pp	$0.93	2.1m	5.77k
2	Xiaomi: MiMo-V2.6-Flash Pareto	79.3%	--	$0.020	6.9m	22.1k
3	Anthropic: Claude Fable 5.1	79.2%	±1.3pp	$0.72	1.9m	6.9k