compass.utilities.costs.LLM_COST_REGISTRY

LLM_COST_REGISTRY = {'gpt-4.1': {'prompt': 2, 'response': 8}, 'gpt-4.1-mini': {'prompt': 0.4, 'response': 1.6}, 'gpt-4.1-nano': {'prompt': 0.1, 'response': 0.4}, 'gpt-4.5': {'prompt': 75, 'response': 150}, 'gpt-4o': {'prompt': 2.5, 'response': 10}, 'gpt-4o-mini': {'prompt': 0.15, 'response': 0.6}, 'gpt-5': {'prompt': 1.25, 'response': 10}, 'gpt-5-chat-latest': {'prompt': 1.25, 'response': 10}, 'gpt-5-mini': {'prompt': 0.25, 'response': 2}, 'gpt-5-nano': {'prompt': 0.05, 'response': 0.4}, 'gpt-5.4': {'prompt': 2.5, 'response': 15}, 'gpt-5.4-mini': {'prompt': 0.75, 'response': 4.5}, 'gpt-5.4-nano': {'prompt': 0.2, 'response': 1.25}, 'gpt-5.5': {'prompt': 5, 'response': 30}, 'gpt-5.6-luna': {'prompt': 0.2, 'response': 1.2}, 'gpt-5.6-luna-aws': {'prompt': 0.22, 'response': 1.32}, 'gpt-5.6-luna-azure': {'prompt': 0.2, 'response': 1.2}, 'gpt-5.6-sol': {'prompt': 5, 'response': 30}, 'gpt-5.6-terra': {'prompt': 2, 'response': 12}, 'gpt-6-astra': {'prompt': 10, 'response': 50}, 'gpt-6-luna': {'prompt': 0.1, 'response': 0.5}, 'gpt-6-sol': {'prompt': 2, 'response': 10}, 'o1': {'prompt': 15, 'response': 60}, 'o3-mini': {'prompt': 1.1, 'response': 4.4}, 'text-embedding-ada-002': {'prompt': 0.1}}

LLM Costs registry

The registry maps model names to a dictionary that contains the cost (in $/million tokens) for both prompt and response tokens.