mirascope 1.19.0__py3-none-any.whl → 1.20.0__py3-none-any.whl

mirascope/core/costs/_mistral_calculate_cost.py

@@ -0,0 +1,64 @@
+"""Calculate the cost of a completion using the Mistral API."""
+
+from ..base.types import CostMetadata
+
+
+def calculate_cost(
+    metadata: CostMetadata,
+    model: str = "open-mistral-7b",
+) -> float | None:
+    """Calculate the cost of a completion using the Mistral API.
+
+    https://mistral.ai/technology/#pricing
+
+    Model                     Input               Cached     Output
+    mistral-large-latest      $2/1M tokens                   $6/1M tokens
+    pixtral-large-latest      $2/1M tokens                   $6/1M tokens
+    mistral-small-latest      $0.1/1M tokens                 $0.3/1M tokens
+    mistral-saba-latest       $0.2/1M tokens                 $0.6/1M tokens
+    codestral-latest          $0.3/1M tokens                 $0.9/1M tokens
+    ministral-8b-latest       $0.1/1M tokens                 $0.1/1M tokens
+    ministral-3b-latest       $0.04/1M tokens                $0.04/1M tokens
+    mistral-embed             $0.1/1M tokens                 -
+    mistral-moderation-latest $0.1/1M tokens                 -
+    open-mistral-nemo         $0.3/1M tokens                 $0.3/1M tokens
+    open-mistral-7b           $0.25/1M tokens                $0.25/1M tokens
+    open-mixtral-8x7b         $0.7/1M tokens                 $0.7/1M tokens
+    open-mixtral-8x22b        $2/1M tokens                   $6/1M tokens
+    """
+    pricing = {
+        "mistral-large-latest": {"prompt": 0.000_002, "completion": 0.000_006},
+        "pixtral-large-latest": {"prompt": 0.000_002, "completion": 0.000_006},
+        "mistral-small-latest": {"prompt": 0.000_000_1, "completion": 0.000_000_3},
+        "mistral-saba-latest": {"prompt": 0.000_000_2, "completion": 0.000_000_6},
+        "codestral-latest": {"prompt": 0.000_000_3, "completion": 0.000_000_9},
+        "ministral-8b-latest": {"prompt": 0.000_000_1, "completion": 0.000_000_1},
+        "ministral-3b-latest": {"prompt": 0.000_000_04, "completion": 0.000_000_04},
+        "mistral-embed": {"prompt": 0.000_000_1, "completion": 0},
+        "mistral-moderation-latest": {"prompt": 0.000_000_1, "completion": 0},
+        "open-mistral-nemo": {"prompt": 0.000_000_3, "completion": 0.000_000_3},
+        "open-mistral-nemo-2407": {"prompt": 0.000_000_3, "completion": 0.000_000_3},
+        "open-mistral-7b": {"prompt": 0.000_000_25, "completion": 0.000_000_25},
+        "open-mixtral-8x7b": {"prompt": 0.000_000_7, "completion": 0.000_000_7},
+        "open-mixtral-8x22b": {"prompt": 0.000_002, "completion": 0.000_006},
+        "mistral-large-2407": {"prompt": 0.000_003, "completion": 0.000_009},
+        "mistral-medium-latest": {"prompt": 0.000_002_75, "completion": 0.000_008_1},
+        "pixtral-12b-2409": {"prompt": 0.000_002, "completion": 0.000_006},
+    }
+
+    if metadata.input_tokens is None or metadata.output_tokens is None:
+        return None
+
+    try:
+        model_pricing = pricing[model]
+    except KeyError:
+        return None
+
+    # Calculate cost for text tokens
+    prompt_cost = metadata.input_tokens * model_pricing["prompt"]
+    completion_cost = metadata.output_tokens * model_pricing["completion"]
+    total_cost = prompt_cost + completion_cost
+
+    # Image tokens is included in the cost
+
+    return total_cost

mirascope/core/costs/_openai_calculate_cost.py

@@ -0,0 +1,416 @@
+"""Calculate the cost of a completion using the OpenAI API."""
+
+from ..base.types import CostMetadata
+
+# Constants for image token calculation
+LOW_DETAIL_IMAGE_TOKENS = 85
+HIGH_DETAIL_TILE_TOKENS = 170
+HIGH_DETAIL_BASE_TOKENS = 85
+TILE_SIZE = 512
+
+
+def _calculate_image_tokens(metadata: CostMetadata) -> int | None:
+    """Calculate tokens used by images based on their size and detail level.
+
+    https://platform.openai.com/docs/guides/vision
+    Following OpenAI's pricing structure:
+    - Low detail: 85 tokens per image
+    - High detail: 85 tokens base + 170 tokens per 512px tile
+      (after scaling to fit within 2048x2048 and making shortest side 768px)
+    """
+    if not metadata.images or not metadata.images:
+        return 0
+
+    total_image_tokens = 0
+
+    for img in metadata.images:
+        if not img.width or not img.height:
+            continue
+
+        # If image already has precalculated tokens, use those
+        if img.tokens is not None:
+            total_image_tokens += img.tokens
+            continue
+
+        if img.detail is not None and img.detail != "auto":
+            detail = img.detail
+        else:
+            # Default to high detail for auto
+            # We can't determine detail level from image alone
+            detail = "high"
+        if detail == "low":
+            # Low detail is a fixed cost regardless of size
+            total_image_tokens += LOW_DETAIL_IMAGE_TOKENS
+        else:
+            # High detail calculation
+
+            # Scale to fit within 2048x2048 square
+            width, height = img.width, img.height
+            if width > 2048 or height > 2048:
+                aspect_ratio = width / height
+                if width > height:
+                    width = 2048
+                    height = int(width / aspect_ratio)
+                else:
+                    height = 2048
+                    width = int(height * aspect_ratio)
+
+            # Scale so shortest side is 768px
+            if min(width, height) > 768:
+                if width < height:
+                    scale_factor = 768 / width
+                    width = 768
+                    height = int(height * scale_factor)
+                else:
+                    scale_factor = 768 / height
+                    height = 768
+                    width = int(width * scale_factor)
+
+            # Calculate number of 512px tiles needed
+            tiles_x = (width + TILE_SIZE - 1) // TILE_SIZE
+            tiles_y = (height + TILE_SIZE - 1) // TILE_SIZE
+            num_tiles = tiles_x * tiles_y
+
+            # Calculate token cost
+            image_tokens = (
+                HIGH_DETAIL_TILE_TOKENS * num_tiles
+            ) + HIGH_DETAIL_BASE_TOKENS
+            total_image_tokens += image_tokens
+
+    return total_image_tokens
+
+
+def calculate_cost(
+    metadata: CostMetadata,
+    model: str = "gpt-3.5-turbo-16k",
+) -> float | None:
+    """Calculate the cost of a completion using the OpenAI API.
+
+    https://openai.com/pricing
+
+    Model                                     Input               Cached               Output
+    gpt-4.5-preview                           $75.00 / 1M tokens  $37.50 / 1M tokens   $150.00 / 1M tokens
+    gpt-4.5-preview-2025-02-27                $75.00 / 1M tokens  $37.50 / 1M tokens   $150.00 / 1M tokens
+    gpt-4o                                    $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-2024-11-20                         $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-2024-08-06                         $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-2024-05-13                         $5.00  / 1M tokens                       $15.00 / 1M tokens
+    gpt-4o-audio-preview                      $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-audio-preview-2024-12-17           $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-audio-preview-2024-10-01           $2.50  / 1M tokens  $1.25 / 1M tokens    $10.00 / 1M tokens
+    gpt-4o-realtime-preview                   $5.00  / 1M tokens  $2.50 / 1M tokens    $20.00 / 1M tokens
+    gpt-4o-realtime-preview-2024-12-17        $5.00  / 1M tokens  $2.50 / 1M tokens    $20.00 / 1M tokens
+    gpt-4o-realtime-preview-2024-10-01        $5.00  / 1M tokens  $2.50 / 1M tokens    $20.00 / 1M tokens
+    gpt-4o-mini                               $0.150 / 1M tokens  $0.075 / 1M tokens   $0.600 / 1M tokens
+    gpt-4o-mini-2024-07-18                    $0.150 / 1M tokens  $0.075 / 1M tokens   $0.600 / 1M tokens
+    gpt-4o-mini-audio-preview                 $0.150 / 1M tokens  $0.075 / 1M tokens   $0.600 / 1M tokens
+    gpt-4o-mini-audio-preview-2024-12-17      $0.150 / 1M tokens  $0.075 / 1M tokens   $0.600 / 1M tokens
+    gpt-4o-mini-realtime-preview              $0.60  / 1M tokens  $0.30 / 1M tokens    $2.40  / 1M tokens
+    gpt-4o-mini-realtime-preview-2024-12-17   $0.60  / 1M tokens  $0.30 / 1M tokens    $2.40  / 1M tokens
+    o1                                        $15.00 / 1M tokens  $7.50 / 1M tokens    $60.00 / 1M tokens
+    o1-2024-12-17                             $15.00 / 1M tokens  $7.50 / 1M tokens    $60.00 / 1M tokens
+    o1-preview                                $15.00 / 1M tokens  $7.50 / 1M tokens    $60.00 / 1M tokens
+    o1-preview-2024-09-12                     $15.00 / 1M tokens  $7.50 / 1M tokens    $60.00 / 1M tokens
+    o3-mini                                   $1.10  / 1M tokens  $0.55 / 1M tokens    $4.40  / 1M tokens
+    o3-mini-2025-01-31                        $1.10  / 1M tokens  $0.55 / 1M tokens    $4.40  / 1M tokens
+    o1-mini                                   $1.10  / 1M tokens  $0.55 / 1M tokens    $4.40  / 1M tokens
+    o1-mini-2024-09-12                        $1.10  / 1M tokens  $0.55 / 1M tokens    $4.40  / 1M tokens
+    chatgpt-4o-latest                         $5.00  / 1M tokens                       $15.00 / 1M tokens
+    gpt-4-turbo                               $10.00 / 1M tokens                       $30.00 / 1M tokens
+    gpt-4-turbo-2024-04-09                    $10.00 / 1M tokens                       $30.00 / 1M tokens
+    gpt-3.5-turbo-0125                        $0.50  / 1M tokens                       $1.50  / 1M tokens
+    gpt-3.5-turbo-1106                        $1.00  / 1M tokens                       $2.00  / 1M tokens
+    gpt-4-0125-preview                        $10.00 / 1M tokens                       $30.00 / 1M tokens
+    gpt-4-1106-preview                        $10.00 / 1M tokens                       $30.00 / 1M tokens
+    gpt-4-vision-preview                      $10.00 / 1M tokens                       $30.00 / 1M tokens
+    gpt-4                                     $30.00 / 1M tokens                       $60.00 / 1M tokens
+    gpt-4-32k                                 $60.00 / 1M tokens                       $120.00 / 1M tokens
+    text-embedding-3-small                    $0.02  / 1M tokens
+    text-embedding-3-large                    $0.13  / 1M tokens
+    text-embedding-ada-002                    $0.10  / 1M tokens
+    """
+    pricing = {
+        "gpt-4.5-preview": {
+            "prompt": 0.000_075,
+            "cached": 0.000_037_5,
+            "completion": 0.000_15,
+            "batch_prompt": 0.000_037_5,
+            "batch_completion": 0.000_075,
+        },
+        "gpt-4.5-preview-2025-02-27": {
+            "prompt": 0.000_075,
+            "cached": 0.000_037_5,
+            "completion": 0.000_15,
+        },
+        "gpt-4o": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+            "batch_prompt": 0.000_001_25,
+            "batch_completion": 0.000_005,
+        },
+        "gpt-4o-2024-11-20": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+        },
+        "gpt-4o-2024-08-06": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+        },
+        "gpt-4o-2024-05-13": {
+            "prompt": 0.000_005,
+            "cached": 0.000_002_5,
+            "completion": 0.000_015,
+        },
+        "gpt-4o-audio-preview": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+        },
+        "gpt-4o-audio-preview-2024-12-17": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+        },
+        "gpt-4o-audio-preview-2024-10-01": {
+            "prompt": 0.000_002_5,
+            "cached": 0.000_001_25,
+            "completion": 0.000_01,
+        },
+        "gpt-4o-realtime-preview": {
+            "prompt": 0.000_005,
+            "cached": 0.000_002_5,
+            "completion": 0.000_02,
+        },
+        "gpt-4o-realtime-preview-2024-12-17": {
+            "prompt": 0.000_005,
+            "cached": 0.000_002_5,
+            "completion": 0.000_02,
+        },
+        "gpt-4o-realtime-preview-2024-10-01": {
+            "prompt": 0.000_005,
+            "cached": 0.000_002_5,
+            "completion": 0.000_02,
+        },
+        "gpt-4o-mini": {
+            "prompt": 0.000_000_15,
+            "cached": 0.000_000_075,
+            "completion": 0.000_000_6,
+        },
+        "gpt-4o-mini-2024-07-18": {
+            "prompt": 0.000_000_15,
+            "cached": 0.000_000_075,
+            "completion": 0.000_000_6,
+        },
+        "gpt-4o-mini-audio-preview": {
+            "prompt": 0.000_000_15,
+            "cached": 0.000_000_075,
+            "completion": 0.000_000_6,
+        },
+        "gpt-4o-mini-audio-preview-2024-12-17": {
+            "prompt": 0.000_000_15,
+            "cached": 0.000_000_075,
+            "completion": 0.000_000_6,
+        },
+        "gpt-4o-mini-realtime-preview": {
+            "prompt": 0.000_000_6,
+            "cached": 0.000_000_3,
+            "completion": 0.000_002_4,
+        },
+        "gpt-4o-mini-realtime-preview-2024-12-17": {
+            "prompt": 0.000_000_6,
+            "cached": 0.000_000_3,
+            "completion": 0.000_002_4,
+        },
+        "o1": {
+            "prompt": 0.000_015,
+            "cached": 0.000_007_5,
+            "completion": 0.000_06,
+        },
+        "o1-2024-12-17": {
+            "prompt": 0.000_015,
+            "cached": 0.000_007_5,
+            "completion": 0.000_06,
+        },
+        "o1-preview": {
+            "prompt": 0.000_015,
+            "cached": 0.000_007_5,
+            "completion": 0.000_06,
+        },
+        "o1-preview-2024-09-12": {
+            "prompt": 0.000_015,
+            "cached": 0.000_007_5,
+            "completion": 0.000_06,
+        },
+        "o3-mini": {
+            "prompt": 0.000_001_1,
+            "cached": 0.000_000_55,
+            "completion": 0.000_004_4,
+        },
+        "o3-mini-2025-01-31": {
+            "prompt": 0.000_001_1,
+            "cached": 0.000_000_55,
+            "completion": 0.000_004_4,
+        },
+        "o1-mini": {
+            "prompt": 0.000_001_1,
+            "cached": 0.000_000_55,
+            "completion": 0.000_004_4,
+        },
+        "o1-mini-2024-09-12": {
+            "prompt": 0.000_001_1,
+            "cached": 0.000_000_55,
+            "completion": 0.000_004_4,
+        },
+        "chatgpt-4o-latest": {
+            "prompt": 0.000_005,
+            "cached": 0,
+            "completion": 0.000_015,
+        },
+        "gpt-4-turbo": {
+            "prompt": 0.000_01,
+            "cached": 0,
+            "completion": 0.000_03,
+        },
+        "gpt-4-turbo-2024-04-09": {
+            "prompt": 0.000_01,
+            "cached": 0,
+            "completion": 0.000_03,
+        },
+        "gpt-3.5-turbo-0125": {
+            "prompt": 0.000_000_5,
+            "cached": 0,
+            "completion": 0.000_001_5,
+        },
+        "gpt-3.5-turbo-1106": {
+            "prompt": 0.000_001,
+            "cached": 0,
+            "completion": 0.000_002,
+        },
+        "gpt-4-0125-preview": {
+            "prompt": 0.000_01,
+            "cached": 0,
+            "completion": 0.000_03,
+        },
+        "gpt-4-1106-preview": {
+            "prompt": 0.000_01,
+            "cached": 0,
+            "completion": 0.000_03,
+        },
+        "gpt-4-vision-preview": {
+            "prompt": 0.000_01,
+            "cached": 0,
+            "completion": 0.000_03,
+        },
+        "gpt-4": {
+            "prompt": 0.000_03,
+            "cached": 0,
+            "completion": 0.000_06,
+        },
+        "gpt-4-32k": {
+            "prompt": 0.000_06,
+            "cached": 0,
+            "completion": 0.000_12,
+        },
+        "gpt-3.5-turbo-4k": {
+            "prompt": 0.000_015,
+            "cached": 0,
+            "completion": 0.000_02,
+        },
+        "gpt-3.5-turbo-16k": {
+            "prompt": 0.000_003,
+            "cached": 0,
+            "completion": 0.000_004,
+        },
+        "gpt-4-8k": {
+            "prompt": 0.000_03,
+            "cached": 0,
+            "completion": 0.000_06,
+        },
+        "text-embedding-3-small": {
+            "prompt": 0.000_000_02,
+            "cached": 0,
+            "completion": 0,
+            "batch_prompt": 0.000_000_01,
+        },
+        "text-embedding-ada-002": {
+            "prompt": 0.000_000_1,
+            "cached": 0,
+            "completion": 0,
+            "batch_prompt": 0.000_000_05,
+        },
+        "text-embedding-3-large": {
+            "prompt": 0.000_000_13,
+            "cached": 0,
+            "completion": 0,
+            "batch_prompt": 0.000_000_065,
+        },
+    }
+
+    # Audio pricing for audio models (per-minute rates in dollars)
+
+    if metadata.cost is not None:
+        return metadata.cost
+
+    # Audio input/output costs
+    # ChatCompletion.usage has brake down of audio input and output.
+    # The total cost already includes the audio input/output cost.
+
+    # Initialize cached tokens if not provided
+    if metadata.cached_tokens is None:
+        metadata.cached_tokens = 0
+
+    # Try to get model pricing
+    try:
+        model_pricing = pricing[model]
+    except KeyError:
+        return None
+
+    image_tokens = _calculate_image_tokens(metadata) or 0
+
+    input_tokens = (metadata.input_tokens or 0) + image_tokens
+
+    # Calculate costs for each component
+    prompt_cost = input_tokens * model_pricing["prompt"]
+    cached_cost = metadata.cached_tokens * model_pricing["cached"]
+    completion_cost = (metadata.output_tokens or 0) * model_pricing["completion"]
+
+    # Special handling for embedding models (only input tokens matter)
+    if "embedding" in model:
+        total_cost = prompt_cost
+    else:
+        total_cost = prompt_cost + cached_cost + completion_cost
+
+    # Apply batch discounts if applicable
+    if metadata.batch_mode:
+        # Based on the OpenAI pricing table, batch mode typically provides
+        # approximately 50% discount for both input and output tokens
+        if "embedding" in model.lower():
+            # Embedding models have specific batch pricing
+            if model == "text-embedding-3-small":
+                prompt_cost = (
+                    input_tokens * 0.000_000_01
+                )  # $0.01 per 1M tokens in batch mode
+            elif model == "text-embedding-3-large":
+                prompt_cost = (
+                    input_tokens * 0.000_000_065
+                )  # $0.065 per 1M tokens in batch mode
+            elif model == "text-embedding-ada-002":
+                prompt_cost = (
+                    input_tokens * 0.000_000_05
+                )  # $0.05 per 1M tokens in batch mode
+        else:
+            # For LLM models, typically 50% discount
+            prompt_cost *= 0.5
+            cached_cost *= 0.5
+            completion_cost *= 0.5
+
+        # Recalculate total cost with batch pricing
+        if "embedding" in model:
+            total_cost = prompt_cost
+        else:
+            total_cost = prompt_cost + cached_cost + completion_cost
+
+    return total_cost

mirascope/core/{vertex/_utils/_calculate_cost.py → costs/_vertex_calculate_cost.py}

@@ -1,12 +1,11 @@
 """Calculate the cost of a completion using the Vertex AI Gemini API, considering context window size."""
 
+from ..base.types import CostMetadata
+
 
 def calculate_cost(
-    input_chars: int | float | None,
-    cached_chars: int | float | None,
-    output_chars: int | float | None,
+    metadata: CostMetadata,
     model: str = "gemini-1.5-pro",
-    context_length: int = 0,
 ) -> float | None:
     """Calculate the cost of a completion using the Vertex AI Gemini API.
 
@@ -19,6 +18,8 @@ def calculate_cost(
 
     Note: Prices are per 1k characters. Gemini 1.0 Pro only supports up to 32K context window.
     """
+
+    context_length = metadata.context_length or 0
     pricing = {
         "gemini-1.5-flash": {
             "prompt_short": 0.000_018_75,
@@ -40,7 +41,7 @@ def calculate_cost(
         },
     }
 
-    if input_chars is None or output_chars is None:
+    if metadata.input_tokens is None or metadata.output_tokens is None:
         return None
 
     try:
@@ -59,8 +60,8 @@ def calculate_cost(
         "completion_long" if use_long_context else "completion_short"
     ]
 
-    prompt_cost = (input_chars / 1000) * prompt_price
-    completion_cost = (output_chars / 1000) * completion_price
+    prompt_cost = (metadata.input_tokens / 1000) * prompt_price
+    completion_cost = (metadata.output_tokens / 1000) * completion_price
     total_cost = prompt_cost + completion_cost
 
     return total_cost

mirascope/core/{xai/_utils/_calculate_cost.py → costs/_xai_calculate_cost.py}

@@ -1,10 +1,10 @@
 """Calculate the cost of a Grok API call."""
 
+from ..base.types import CostMetadata
+
 
 def calculate_cost(
-    input_tokens: int | float | None,
-    cached_tokens: int | float | None,
-    output_tokens: int | float | None,
+    metadata: CostMetadata,
     model: str,
 ) -> float | None:
     """Calculate the cost of an xAI Grok API call.
@@ -81,11 +81,11 @@ def calculate_cost(
         },
     }
 
-    if input_tokens is None or output_tokens is None:
+    if metadata.input_tokens is None or metadata.output_tokens is None:
         return None
 
-    if cached_tokens is None:
-        cached_tokens = 0
+    if metadata.cached_tokens is None:
+        metadata.cached_tokens = 0
 
     try:
         model_pricing = pricing[model]
@@ -96,9 +96,9 @@ def calculate_cost(
     cached_price = model_pricing["cached"]
     completion_price = model_pricing["completion"]
 
-    prompt_cost = input_tokens * prompt_price
-    cached_cost = cached_tokens * cached_price
-    completion_cost = output_tokens * completion_price
+    prompt_cost = metadata.input_tokens * prompt_price
+    cached_cost = metadata.cached_tokens * cached_price
+    completion_cost = metadata.output_tokens * completion_price
     total_cost = prompt_cost + cached_cost + completion_cost
 
     return total_cost

mirascope/core/costs/calculate_cost.py

@@ -0,0 +1,86 @@
+"""Cost calculation utilities for LLM API calls."""
+
+from __future__ import annotations
+
+from ..base.types import CostMetadata, Provider
+from ._anthropic_calculate_cost import (
+    calculate_cost as anthropic_calculate_cost,
+)
+from ._azure_calculate_cost import calculate_cost as azure_calculate_cost
+from ._bedrock_calculate_cost import calculate_cost as bedrock_calculate_cost
+from ._cohere_calculate_cost import calculate_cost as cohere_calculate_cost
+from ._gemini_calculate_cost import calculate_cost as gemini_calculate_cost
+from ._google_calculate_cost import calculate_cost as google_calculate_cost
+from ._groq_calculate_cost import calculate_cost as groq_calculate_cost
+from ._litellm_calculate_cost import calculate_cost as litellm_calculate_cost
+from ._mistral_calculate_cost import calculate_cost as mistral_calculate_cost
+from ._openai_calculate_cost import calculate_cost as openai_calculate_cost
+from ._vertex_calculate_cost import calculate_cost as vertex_calculate_cost
+from ._xai_calculate_cost import calculate_cost as xai_calculate_cost
+
+
+def calculate_cost(
+    provider: Provider,
+    model: str,
+    metadata: CostMetadata | None = None,
+) -> float | None:
+    """Calculate the cost for an LLM API call.
+
+    This function routes to the appropriate provider-specific cost calculation function,
+    preserving existing behavior while providing a unified interface.
+
+    Args:
+        provider: The LLM provider (e.g., "openai", "anthropic")
+        model: The model name (e.g., "gpt-4", "claude-3-opus")
+        metadata: Additional metadata required for cost calculation
+
+    Returns:
+        The calculated cost in USD or None if unable to calculate
+    """
+
+    # Initialize empty metadata if none provided
+    if metadata is None:
+        metadata = CostMetadata()
+
+    # Set default values
+    if metadata.cached_tokens is None:
+        metadata.cached_tokens = 0
+
+    # Route to provider-specific implementations
+    if provider == "openai":
+        return openai_calculate_cost(metadata, model)
+
+    elif provider == "anthropic":
+        return anthropic_calculate_cost(metadata, model)
+
+    elif provider == "azure":
+        return azure_calculate_cost(metadata, model)
+
+    elif provider == "bedrock":
+        return bedrock_calculate_cost(metadata, model)
+
+    elif provider == "cohere":
+        return cohere_calculate_cost(metadata, model)
+
+    elif provider == "gemini":
+        return gemini_calculate_cost(metadata, model)
+
+    elif provider == "google":
+        return google_calculate_cost(metadata, model)
+
+    elif provider == "groq":
+        return groq_calculate_cost(metadata, model)
+
+    elif provider == "mistral":
+        return mistral_calculate_cost(metadata, model)
+
+    elif provider == "vertex":
+        return vertex_calculate_cost(metadata, model)
+
+    elif provider == "xai":
+        return xai_calculate_cost(metadata, model)
+
+    elif provider == "litellm":
+        return litellm_calculate_cost(metadata, model)
+    else:
+        raise ValueError(f"Unsupported provider: {provider}")

mirascope/core/gemini/_utils/__init__.py

@@ -1,13 +1,11 @@
 """Gemini utilities for decorator factories."""
 
-from ._calculate_cost import calculate_cost
 from ._convert_message_params import convert_message_params
 from ._get_json_output import get_json_output
 from ._handle_stream import handle_stream, handle_stream_async
 from ._setup_call import setup_call
 
 __all__ = [
-    "calculate_cost",
     "convert_message_params",
     "get_json_output",
     "handle_stream",