diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index b08b72a5434..0c55e3e45bf 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -86905,5 +86905,123 @@ "mode": "rerank", "output_cost_per_token": 0.0, "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'" + }, + "azure/us/gpt-realtime-2.1": { + "cache_read_input_audio_token_cost": 4.4e-07, + "cache_read_input_image_token_cost": 5.5e-07, + "cache_read_input_token_cost": 4.4e-07, + "deprecation_date": "2027-06-25", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 3.52e-05, + "input_cost_per_image_token": 5.5e-06, + "input_cost_per_token": 4.4e-06, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 7.04e-05, + "output_cost_per_token": 2.64e-05, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_system_messages": true, + "supports_tool_choice": true + }, + "azure/us/gpt-realtime-2.1-mini": { + "cache_read_input_audio_token_cost": 3.3e-07, + "cache_read_input_image_token_cost": 8.8e-08, + "cache_read_input_token_cost": 6.6e-08, + "deprecation_date": "2027-06-25", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 1.1e-05, + "input_cost_per_image_token": 8.8e-07, + "input_cost_per_token": 6.6e-07, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 2.2e-05, + "output_cost_per_token": 2.64e-06, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_system_messages": true, + "supports_tool_choice": true + }, + "azure/us/gpt-realtime-1.5": { + "cache_read_input_audio_token_cost": 4.4e-07, + "cache_read_input_image_token_cost": 5.5e-07, + "cache_read_input_token_cost": 4.4e-07, + "deprecation_date": "2027-08-24", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 3.52e-05, + "input_cost_per_image_token": 5.5e-06, + "input_cost_per_token": 4.4e-06, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 7.04e-05, + "output_cost_per_token": 1.76e-05, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_system_messages": true, + "supports_tool_choice": true } } diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index b08b72a5434..0c55e3e45bf 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -86905,5 +86905,123 @@ "mode": "rerank", "output_cost_per_token": 0.0, "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'" + }, + "azure/us/gpt-realtime-2.1": { + "cache_read_input_audio_token_cost": 4.4e-07, + "cache_read_input_image_token_cost": 5.5e-07, + "cache_read_input_token_cost": 4.4e-07, + "deprecation_date": "2027-06-25", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 3.52e-05, + "input_cost_per_image_token": 5.5e-06, + "input_cost_per_token": 4.4e-06, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 7.04e-05, + "output_cost_per_token": 2.64e-05, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_system_messages": true, + "supports_tool_choice": true + }, + "azure/us/gpt-realtime-2.1-mini": { + "cache_read_input_audio_token_cost": 3.3e-07, + "cache_read_input_image_token_cost": 8.8e-08, + "cache_read_input_token_cost": 6.6e-08, + "deprecation_date": "2027-06-25", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 1.1e-05, + "input_cost_per_image_token": 8.8e-07, + "input_cost_per_token": 6.6e-07, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 2.2e-05, + "output_cost_per_token": 2.64e-06, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_system_messages": true, + "supports_tool_choice": true + }, + "azure/us/gpt-realtime-1.5": { + "cache_read_input_audio_token_cost": 4.4e-07, + "cache_read_input_image_token_cost": 5.5e-07, + "cache_read_input_token_cost": 4.4e-07, + "deprecation_date": "2027-08-24", + "deprecation_source": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule", + "input_cost_per_audio_token": 3.52e-05, + "input_cost_per_image_token": 5.5e-06, + "input_cost_per_token": 4.4e-06, + "litellm_provider": "azure", + "max_input_tokens": 32000, + "max_output_tokens": 4096, + "mode": "realtime", + "output_cost_per_audio_token": 7.04e-05, + "output_cost_per_token": 1.76e-05, + "provider_specific_entry": { + "realtime_ga_only": 1 + }, + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", + "supported_endpoints": [ + "/v1/realtime" + ], + "supported_modalities": [ + "text", + "image", + "audio" + ], + "supported_output_modalities": [ + "text", + "audio" + ], + "supports_audio_input": true, + "supports_audio_output": true, + "supports_function_calling": true, + "supports_parallel_function_calling": true, + "supports_system_messages": true, + "supports_tool_choice": true } }