> ## Documentation Index
> Fetch the complete documentation index at: https://docs.galtea.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Model

> A representation of a LLM Model with cost information to calculate cost estimations

> [!WARNING]
>
> You are reading the docs of Galtea 5.13. [Read this page in the latest version](https://docs.galtea.ai/concepts/model.md).

## What is a Model?

A model in Galtea represents a Large Language Model (LLM) configuration with associated cost information. This allows the platform to track, estimate, and report on the costs of using different LLMs across your products and evaluations.

> [!NOTE]
>
> Models are organization-wide and can be referenced across multiple products to ensure consistent cost tracking.

You can **create**, view and manage your models on the [Galtea dashboard](https://platform.galtea.ai/) or programmatically via the [SDK](https://docs.galtea.ai/v5.13/sdk/api/model/service.md).

## Use Cases for Models

Models in Galtea are primarily used for:

> [card] Cost Tracking
>
> Monitor spending on AI models across different products and versions

> [card] Budget Planning
>
> Project future costs based on token usage patterns

> [card] Cost Optimization
>
> Compare costs between different model providers and configurations

> [card] Financial Reporting
>
> Generate accurate reports on LLM usage expenses

## Models vs. Evaluator Models

It's important to distinguish between a **Model** and an **Evaluator Model** in Galtea:

* **Model (This Page):** Represents the AI model your product is using. Its primary purpose in Galtea is for **cost estimation and tracking**. You link a Version to a Model to calculate the cost of its traces. A version with a [model fact](https://docs.galtea.ai/v5.13/concepts/product/version.md#model-facts) such as `model: azure/gpt-4o` is linked to the model with that name, ignoring case, when no model is set.

* **Evaluator Model:** This is an LLM used by Galtea to perform "LLM-as-a-judge" evaluations for non-deterministic [Metrics](https://docs.galtea.ai/v5.13/concepts/metric.md). You select an Evaluator Model when creating a custom metric with a `judge_prompt`.

While both are LLMs, their roles within the Galtea platform are distinct: one is being tested, and the other is doing the testing.

## Model Properties

When creating a model in Galtea, you'll need to provide the following information:

**Name** `Text` *required*

The name of the model. This should be descriptive and indicate the provider and version.
**Example**: "GPT-4 Turbo" or "Claude 3 Haiku"

**Cost per Input Token** `Number`

The cost in dollars per input token. This is the rate charged by the provider for tokens in your prompts.
**Example**: 0.00001 (representing \$0.00001 per token)

**Cost per Output Token** `Number`

The cost in dollars per output token. This is the rate charged by the provider for tokens in the model's responses.
**Example**: 0.00003 (representing \$0.00003 per token)

**Cost per Cache Read Input Token** `Number`

The cost in dollars per cached input token. Some providers offer reduced rates for cached requests.
**Example**: 0.000005 (representing \$0.000005 per token)

**Cost per Cache Creation Input Token** `Number`

The cost in dollars per input token when creating a cache entry.
**Example**: 0.000005

**Cost per Input Token (Above 128k)** `Number`

The cost in dollars per input token for contexts larger than 128k tokens, if applicable.
**Example**: 0.00002

**Cost per Output Token (Above 128k)** `Number`

The cost in dollars per output token for contexts larger than 128k tokens, if applicable.
**Example**: 0.00006

**Tokenizer Provider** `Text`

The provider of the tokenizer used by the model. This is important for accurate token counting and cost estimation.
Right now, the only supported tokenizers are:
**Examples**:

* "OpenAI" (uses [tiktoken](https://github.com/openai/tiktoken))
* "Anthropic" (uses [anthropic-tokenizer](https://lunary.ai/anthropic-tokenizer))

**Source** `Text`

The provider or source of the pricing model. This can be a URL to the model's pricing page or documentation.
For instance: [https://openai.com/api/pricing/](https://openai.com/api/pricing/)

## Related

> [card] [Concepts overview](https://docs.galtea.ai/v5.13/concepts/overview.md)
>
> How Galtea's concepts connect — diagram + per-entity quick reference.

> [card] [Model Service SDK](https://docs.galtea.ai/v5.13/sdk/api/model/service.md)
>
> Manage models using the Python SDK

> [card] [Version](https://docs.galtea.ai/v5.13/concepts/product/version.md)
>
> Link a model to a version for cost tracking.

> [card] [Metrics](https://docs.galtea.ai/v5.13/concepts/metric.md)
>
> Select evaluator models when creating custom metrics.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.