diff --git a/pricing_table.md b/pricing_table.md index 7ec0e5e..03ae786 100644 --- a/pricing_table.md +++ b/pricing_table.md @@ -1023,14 +1023,31 @@ | azure/gpt-4o-mini-tts | $2.5 | $10 | nan | nan | | azure_ai/embed-v-4-0 | $0.12 | $0 | 128,000 | nan | | eu.anthropic.claude-3-7-sonnet-20250219-v1:0 | $3 | $15 | 200,000 | 8192 | +| xai/grok-3 | $3 | $15 | 131,072 | 131072 | | groq/llama-guard-3-8b | $0.2 | $0.2 | 8,192 | 8192 | | groq/meta-llama/llama-4-scout-17b-16e-instruct | $0.11 | $0.34 | 131,072 | 8192 | | groq/meta-llama/llama-4-maverick-17b-128e-instruct | $0.2 | $0.6 | 131,072 | 8192 | | groq/mistral-saba-24b | $0.79 | $0.79 | 32,000 | 32000 | | groq/qwen-qwq-32b | $0.29 | $0.39 | 128,000 | 128000 | | groq/playai-tts | -- | -- | 10,000 | 10000 | -| featherless_ai/featherless-ai/Qwerky-72B | -- | -- | 32,768 | 4096 | -| featherless_ai/featherless-ai/Qwerky-QwQ-32B | -- | -- | 32,768 | 4096 | +| claude-opus-4-20250514 | $15 | $75 | 200,000 | 128000 | +| claude-sonnet-4-20250514 | $3 | $15 | 200,000 | 128000 | +| gemini/gemini-2.5-flash-preview-tts | $0.15 | $0.6 | 1,048,576 | 65535 | +| gemini/gemini-2.5-flash-preview-05-20 | $0.15 | $0.6 | 1,048,576 | 65535 | +| gemini-2.5-flash-preview-05-20 | $0.15 | $0.6 | 1,048,576 | 65535 | +| gemini-2.0-flash-preview-image-generation | $0.1 | $0.4 | 1,048,576 | 8192 | +| gemini-2.5-pro-preview-tts | $1.25 | $10 | 1,048,576 | 65535 | +| gemini/gemini-2.0-flash-preview-image-generation | $0.1 | $0.4 | 1,048,576 | 8192 | +| gemini/gemini-2.5-pro-preview-tts | $1.25 | $10 | 1,048,576 | 65535 | +| vertex_ai/claude-opus-4@20250514 | $15 | $75 | 200,000 | 128000 | +| vertex_ai/claude-sonnet-4@20250514 | $3 | $15 | 200,000 | 128000 | +| anthropic.claude-opus-4-20250514-v1:0 | $15 | $75 | 200,000 | 128000 | +| anthropic.claude-sonnet-4-20250514-v1:0 | $3 | $15 | 200,000 | 128000 | +| us.anthropic.claude-opus-4-20250514-v1:0 | $15 | $75 | 200,000 | 128000 | +| us.anthropic.claude-sonnet-4-20250514-v1:0 | $3 | $15 | 200,000 | 128000 | +| eu.anthropic.claude-opus-4-20250514-v1:0 | $15 | $75 | 200,000 | 128000 | +| eu.anthropic.claude-sonnet-4-20250514-v1:0 | $3 | $15 | 200,000 | 128000 | +| databricks/databricks-llama-4-maverick | $5 | $15 | 128,000 | 128000 | | sambanova/Llama-4-Maverick-17B-128E-Instruct | $0.63 | $1.8 | 131,072 | 131072 | | sambanova/Llama-4-Scout-17B-16E-Instruct | $0.4 | $0.7 | 8,192 | 8192 | | sambanova/Meta-Llama-Guard-3-8B | $0.3 | $0.3 | 16,384 | 16384 | @@ -1040,11 +1057,5 @@ | sambanova/DeepSeek-R1-Distill-Llama-70B | $0.7 | $1.4 | 131,072 | 131072 | | sambanova/DeepSeek-R1 | $5 | $7 | 32,768 | 32768 | | sambanova/DeepSeek-V3-0324 | $3 | $4.5 | 32,768 | 32768 | -| xai/grok-3 | $3 | $15 | 131,072 | 131072 | -| gemini/gemini-2.5-flash-preview-tts | $0.15 | $0.6 | 1,048,576 | 65535 | -| gemini/gemini-2.5-flash-preview-05-20 | $0.15 | $0.6 | 1,048,576 | 65535 | -| gemini-2.5-flash-preview-05-20 | $0.15 | $0.6 | 1,048,576 | 65535 | -| gemini-2.0-flash-preview-image-generation | $0.1 | $0.4 | 1,048,576 | 8192 | -| gemini-2.5-pro-preview-tts | $1.25 | $10 | 1,048,576 | 65535 | -| gemini/gemini-2.0-flash-preview-image-generation | $0.1 | $0.4 | 1,048,576 | 8192 | -| gemini/gemini-2.5-pro-preview-tts | $1.25 | $10 | 1,048,576 | 65535 | \ No newline at end of file +| featherless_ai/featherless-ai/Qwerky-72B | -- | -- | 32,768 | 4096 | +| featherless_ai/featherless-ai/Qwerky-QwQ-32B | -- | -- | 32,768 | 4096 | diff --git a/tokencost/model_prices.json b/tokencost/model_prices.json index 40ea87f..772902c 100644 --- a/tokencost/model_prices.json +++ b/tokencost/model_prices.json @@ -13306,6 +13306,19 @@ "supports_tool_choice": true, "supports_reasoning": true }, + "xai/grok-3": { + "max_tokens": 131072, + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "litellm_provider": "xai", + "mode": "chat", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_response_schema": false, + "source": "https://x.ai/api#pricing" + }, "groq/llama-guard-3-8b": { "max_tokens": 8192, "max_input_tokens": 8192, @@ -13369,142 +13382,57 @@ "litellm_provider": "groq", "mode": "audio_speech" }, - "featherless_ai/featherless-ai/Qwerky-72B": { - "max_tokens": 32768, - "max_input_tokens": 32768, - "max_output_tokens": 4096, - "litellm_provider": "featherless_ai", - "mode": "chat" - }, - "featherless_ai/featherless-ai/Qwerky-QwQ-32B": { - "max_tokens": 32768, - "max_input_tokens": 32768, - "max_output_tokens": 4096, - "litellm_provider": "featherless_ai", - "mode": "chat" - }, - "sambanova/Llama-4-Maverick-17B-128E-Instruct": { - "max_tokens": 131072, - "max_input_tokens": 131072, - "max_output_tokens": 131072, - "input_cost_per_token": 6.3e-07, - "output_cost_per_token": 1.8e-06, - "litellm_provider": "sambanova", + "claude-opus-4-20250514": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 1.5e-05, + "output_cost_per_token": 7.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 1.875e-05, + "cache_read_input_token_cost": 1.5e-06, + "litellm_provider": "anthropic", "mode": "chat", "supports_function_calling": true, - "supports_tool_choice": true, - "supports_response_schema": true, "supports_vision": true, - "source": "https://cloud.sambanova.ai/plans/pricing", - "metadata": { - "notes": "For vision models, images are converted to 6432 input tokens and are billed at that amount" - } - }, - "sambanova/Llama-4-Scout-17B-16E-Instruct": { - "max_tokens": 8192, - "max_input_tokens": 8192, - "max_output_tokens": 8192, - "input_cost_per_token": 4e-07, - "output_cost_per_token": 7e-07, - "litellm_provider": "sambanova", - "mode": "chat", - "supports_function_calling": true, - "supports_tool_choice": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, "supports_response_schema": true, - "source": "https://cloud.sambanova.ai/plans/pricing", - "metadata": { - "notes": "For vision models, images are converted to 6432 input tokens and are billed at that amount" - } - }, - "sambanova/Meta-Llama-Guard-3-8B": { - "max_tokens": 16384, - "max_input_tokens": 16384, - "max_output_tokens": 16384, - "input_cost_per_token": 3e-07, - "output_cost_per_token": 3e-07, - "litellm_provider": "sambanova", - "mode": "chat", - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/Qwen3-32B": { - "max_tokens": 8192, - "max_input_tokens": 8192, - "max_output_tokens": 8192, - "input_cost_per_token": 4e-07, - "output_cost_per_token": 8e-07, - "litellm_provider": "sambanova", - "supports_function_calling": true, - "supports_tool_choice": true, - "supports_reasoning": true, - "mode": "chat", - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/QwQ-32B": { - "max_tokens": 16384, - "max_input_tokens": 16384, - "max_output_tokens": 16384, - "input_cost_per_token": 5e-07, - "output_cost_per_token": 1e-06, - "litellm_provider": "sambanova", - "mode": "chat", - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/Qwen2-Audio-7B-Instruct": { - "max_tokens": 4096, - "max_input_tokens": 4096, - "max_output_tokens": 4096, - "input_cost_per_token": 5e-07, - "output_cost_per_token": 0.0001, - "litellm_provider": "sambanova", - "mode": "chat", - "supports_audio_input": true, - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/DeepSeek-R1-Distill-Llama-70B": { - "max_tokens": 131072, - "max_input_tokens": 131072, - "max_output_tokens": 131072, - "input_cost_per_token": 7e-07, - "output_cost_per_token": 1.4e-06, - "litellm_provider": "sambanova", - "mode": "chat", - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/DeepSeek-R1": { - "max_tokens": 32768, - "max_input_tokens": 32768, - "max_output_tokens": 32768, - "input_cost_per_token": 5e-06, - "output_cost_per_token": 7e-06, - "litellm_provider": "sambanova", - "mode": "chat", - "source": "https://cloud.sambanova.ai/plans/pricing" - }, - "sambanova/DeepSeek-V3-0324": { - "max_tokens": 32768, - "max_input_tokens": 32768, - "max_output_tokens": 32768, - "input_cost_per_token": 3e-06, - "output_cost_per_token": 4.5e-06, - "litellm_provider": "sambanova", - "mode": "chat", - "supports_function_calling": true, "supports_tool_choice": true, "supports_reasoning": true, - "source": "https://cloud.sambanova.ai/plans/pricing" + "supports_computer_use": true }, - "xai/grok-3": { - "max_tokens": 131072, - "max_input_tokens": 131072, - "max_output_tokens": 131072, + "claude-sonnet-4-20250514": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, "input_cost_per_token": 3e-06, "output_cost_per_token": 1.5e-05, - "litellm_provider": "xai", + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 3.75e-06, + "cache_read_input_token_cost": 3e-07, + "litellm_provider": "anthropic", "mode": "chat", "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, "supports_tool_choice": true, - "supports_response_schema": false, - "source": "https://x.ai/api#pricing" + "supports_reasoning": true, + "supports_computer_use": true }, "gemini/gemini-2.5-flash-preview-tts": { "max_tokens": 65535, @@ -13755,5 +13683,353 @@ "audio" ], "source": "https://ai.google.dev/gemini-api/docs/pricing#gemini-2.5-pro-preview" + }, + "vertex_ai/claude-opus-4@20250514": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 1.5e-05, + "output_cost_per_token": 7.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 1.875e-05, + "cache_read_input_token_cost": 1.5e-06, + "litellm_provider": "vertex_ai-anthropic_models", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "vertex_ai/claude-sonnet-4@20250514": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 3.75e-06, + "cache_read_input_token_cost": 3e-07, + "litellm_provider": "vertex_ai-anthropic_models", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "anthropic.claude-opus-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 1.5e-05, + "output_cost_per_token": 7.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 1.875e-05, + "cache_read_input_token_cost": 1.5e-06, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "anthropic.claude-sonnet-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 3.75e-06, + "cache_read_input_token_cost": 3e-07, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "us.anthropic.claude-opus-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 1.5e-05, + "output_cost_per_token": 7.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 1.875e-05, + "cache_read_input_token_cost": 1.5e-06, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "us.anthropic.claude-sonnet-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 3.75e-06, + "cache_read_input_token_cost": 3e-07, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "eu.anthropic.claude-opus-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 1.5e-05, + "output_cost_per_token": 7.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 1.875e-05, + "cache_read_input_token_cost": 1.5e-06, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "eu.anthropic.claude-sonnet-4-20250514-v1:0": { + "max_tokens": 128000, + "max_input_tokens": 200000, + "max_output_tokens": 128000, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 1.5e-05, + "search_context_cost_per_query": { + "search_context_size_low": 0.01, + "search_context_size_medium": 0.01, + "search_context_size_high": 0.01 + }, + "cache_creation_input_token_cost": 3.75e-06, + "cache_read_input_token_cost": 3e-07, + "litellm_provider": "bedrock_converse", + "mode": "chat", + "supports_function_calling": true, + "supports_vision": true, + "tool_use_system_prompt_tokens": 159, + "supports_assistant_prefill": true, + "supports_pdf_input": true, + "supports_prompt_caching": true, + "supports_response_schema": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "supports_computer_use": true + }, + "databricks/databricks-llama-4-maverick": { + "max_tokens": 128000, + "max_input_tokens": 128000, + "max_output_tokens": 128000, + "input_cost_per_token": 5e-06, + "input_dbu_cost_per_token": 7.143e-05, + "output_cost_per_token": 1.5e-05, + "output_dbu_cost_per_token": 0.00021429, + "litellm_provider": "databricks", + "mode": "chat", + "source": "https://www.databricks.com/product/pricing/foundation-model-serving", + "metadata": { + "notes": "Databricks documentation now provides both DBU costs (_dbu_cost_per_token) and dollar costs(_cost_per_token)." + }, + "supports_tool_choice": true + }, + "sambanova/Llama-4-Maverick-17B-128E-Instruct": { + "max_tokens": 131072, + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "input_cost_per_token": 6.3e-07, + "output_cost_per_token": 1.8e-06, + "litellm_provider": "sambanova", + "mode": "chat", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_response_schema": true, + "supports_vision": true, + "source": "https://cloud.sambanova.ai/plans/pricing", + "metadata": { + "notes": "For vision models, images are converted to 6432 input tokens and are billed at that amount" + } + }, + "sambanova/Llama-4-Scout-17B-16E-Instruct": { + "max_tokens": 8192, + "max_input_tokens": 8192, + "max_output_tokens": 8192, + "input_cost_per_token": 4e-07, + "output_cost_per_token": 7e-07, + "litellm_provider": "sambanova", + "mode": "chat", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_response_schema": true, + "source": "https://cloud.sambanova.ai/plans/pricing", + "metadata": { + "notes": "For vision models, images are converted to 6432 input tokens and are billed at that amount" + } + }, + "sambanova/Meta-Llama-Guard-3-8B": { + "max_tokens": 16384, + "max_input_tokens": 16384, + "max_output_tokens": 16384, + "input_cost_per_token": 3e-07, + "output_cost_per_token": 3e-07, + "litellm_provider": "sambanova", + "mode": "chat", + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/Qwen3-32B": { + "max_tokens": 8192, + "max_input_tokens": 8192, + "max_output_tokens": 8192, + "input_cost_per_token": 4e-07, + "output_cost_per_token": 8e-07, + "litellm_provider": "sambanova", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "mode": "chat", + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/QwQ-32B": { + "max_tokens": 16384, + "max_input_tokens": 16384, + "max_output_tokens": 16384, + "input_cost_per_token": 5e-07, + "output_cost_per_token": 1e-06, + "litellm_provider": "sambanova", + "mode": "chat", + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/Qwen2-Audio-7B-Instruct": { + "max_tokens": 4096, + "max_input_tokens": 4096, + "max_output_tokens": 4096, + "input_cost_per_token": 5e-07, + "output_cost_per_token": 0.0001, + "litellm_provider": "sambanova", + "mode": "chat", + "supports_audio_input": true, + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/DeepSeek-R1-Distill-Llama-70B": { + "max_tokens": 131072, + "max_input_tokens": 131072, + "max_output_tokens": 131072, + "input_cost_per_token": 7e-07, + "output_cost_per_token": 1.4e-06, + "litellm_provider": "sambanova", + "mode": "chat", + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/DeepSeek-R1": { + "max_tokens": 32768, + "max_input_tokens": 32768, + "max_output_tokens": 32768, + "input_cost_per_token": 5e-06, + "output_cost_per_token": 7e-06, + "litellm_provider": "sambanova", + "mode": "chat", + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "sambanova/DeepSeek-V3-0324": { + "max_tokens": 32768, + "max_input_tokens": 32768, + "max_output_tokens": 32768, + "input_cost_per_token": 3e-06, + "output_cost_per_token": 4.5e-06, + "litellm_provider": "sambanova", + "mode": "chat", + "supports_function_calling": true, + "supports_tool_choice": true, + "supports_reasoning": true, + "source": "https://cloud.sambanova.ai/plans/pricing" + }, + "featherless_ai/featherless-ai/Qwerky-72B": { + "max_tokens": 32768, + "max_input_tokens": 32768, + "max_output_tokens": 4096, + "litellm_provider": "featherless_ai", + "mode": "chat" + }, + "featherless_ai/featherless-ai/Qwerky-QwQ-32B": { + "max_tokens": 32768, + "max_input_tokens": 32768, + "max_output_tokens": 4096, + "litellm_provider": "featherless_ai", + "mode": "chat" } } \ No newline at end of file