genai-prices

0.0.56 · active · verified Sat Apr 11

genai-prices is a Python library for calculating the cost of interacting with various Large Language Model (LLM) inference APIs. It aggregates pricing information from major providers like OpenAI, Anthropic, Google, AWS Bedrock, and more. The library is actively maintained by Pydantic, with frequent minor releases (typically weekly or bi-weekly) to keep pricing data and model definitions up-to-date.

Common errors

```
AttributeError: 'NoneType' object has no attribute 'total_price'
```
cause The `calc_price` function returns `None` when it cannot find pricing information for the specified model and/or provider, and the code attempts to access attributes (like `total_price`) on this `None` result without a check.

fix Always check if the result of `calc_price` is not `None` before attempting to access its attributes. Ensure the `model_id` and `provider_id` are correct and present in the library's data, potentially using `genai_prices.list_models()` to verify.
```
ValueError: Uncached text input tokens cannot be negative
```
cause This error occurs when the token usage parameters passed to `calc_price` result in a calculated negative value for uncached input tokens, indicating an inconsistency in how `input_tokens`, `cache_read_tokens`, and `cache_write_tokens` are provided.

fix Ensure that `input_tokens` represents the total number of input tokens, and that `cache_read_tokens` and `cache_write_tokens` (if provided) are subsets of `input_tokens` such that `input_tokens - cache_read_tokens - cache_write_tokens` is not negative. If using `cache_read_tokens` and `cache_write_tokens`, `input_tokens` should generally be greater than or equal to their sum.
```
ModuleNotFoundError: No module named 'genai_prices'
```
cause The `genai-prices` package is not installed in the Python environment, or the environment where the code is being run does not have access to the installed package.

fix Install the library using pip: `pip install genai-prices`. If using a virtual environment, activate it before installation. Ensure your IDE or script execution environment is using the correct Python interpreter where the package is installed.

Warnings

gotcha Model names and pricing are frequently updated. Relying on exact string matches for models can be brittle. It's recommended to keep the library updated to ensure you have the latest model aliases and pricing information.
Fix: Regularly update the `genai-prices` library (`pip install --upgrade genai-prices`) and consider using the `genai-prices models` CLI command or `get_model_price_and_alias` function (if available) to verify model names.
breaking As of v0.0.54, the library introduced 'model lifecycle management' meaning models can be explicitly marked as deprecated or removed. If you rely on a model that is subsequently removed, `get_model_price` will return `None`.
Fix: Implement error handling for `None` returns from `get_model_price` and ensure your application can gracefully handle models being retired. Consult provider documentation for active models or use the library's CLI (`genai-prices models`) to check current status.
gotcha AWS Bedrock model matching logic was loosened in v0.0.55/v0.0.56 to support inference profiles and regional models. This change might alter which specific model is matched if your input string is ambiguous, potentially leading to different price calculations.
Fix: Test your existing model name inputs with the updated library to confirm the correct model is being matched. Use specific and unambiguous model names where possible, or inspect the `ModelInfo` returned to verify the matched provider and model ID.
gotcha Older versions (pre-v0.0.50) might have had incorrect tiered pricing calculations, as a fix for this was included in v0.0.50.
Fix: Upgrade to `genai-prices` v0.0.50 or newer to ensure accurate tiered pricing calculations.
breaking The `get_model_price` function is no longer directly importable from the `genai_prices` package, leading to an `ImportError`.
Fix: Update your code to reflect the new import path or function name for model price retrieval. Consult the official `genai-prices` documentation or release notes for the correct API usage in the installed version.
breaking The `get_model_price` function is no longer directly importable from the top-level `genai_prices` package. It has likely been moved, renamed, or replaced by functions such as `get_model_price_and_alias`.
Fix: Update your imports to reflect the new location or name of the price retrieval function. Consult the latest `genai-prices` documentation or source code, and consider using `get_model_price_and_alias` if it fits your needs.

Install

pip install genai-prices Install latest version

Imports

get_model_price

from genai_prices import get_model_price

TokenUsage
```
from genai_prices import TokenUsage
```
ModelInfo
wrong:
```
from genai_prices.model_info import ModelInfo
```
correct:
```
from genai_prices import ModelInfo
```
Top-level imports are preferred for core components; ModelInfo is exposed directly via `__init__.py`.

Quickstart

This example demonstrates how to use `get_model_price` to calculate the cost for a given model and token usage. It prints the per-1k-token costs and the total cost. If the model is not found, it prints a message.

from genai_prices import get_model_price, TokenUsage
from decimal import Decimal

model_name = "gpt-4o"
usage = TokenUsage(
    input_tokens=1000,
    output_tokens=1000,
)
price_info = get_model_price(model_name, usage)

if price_info:
    print(f"Model: {model_name}")
    print(f"Provider: {price_info.provider}")
    print(f"Input price per 1k tokens: ${price_info.input_cost_per_token_1k}")
    print(f"Output price per 1k tokens: ${price_info.output_cost_per_token_1k}")
    print(f"Total cost: ${price_info.total_cost:.6f}")
else:
    print(f"Price for model '{model_name}' not found.")

view raw JSON →