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

# Models

> Configure Anthropic Claude and OpenAI GPT models for agents

> Define LLM configurations that agents and teams use for inference.

<Tabs>
  <Tab title="Anthropic (Claude)">
    ## `agno/models/anthropic`

    Configures an Anthropic Claude model.

    ### Config

    | Field            | Type            | Required | Default | Description                                                     |
    | ---------------- | --------------- | -------- | ------- | --------------------------------------------------------------- |
    | `id`             | `string`        | **Yes**  | —       | Model identifier (e.g., `claude-sonnet-4-20250514`)             |
    | `api_key`        | `Field[string]` | **Yes**  | —       | Anthropic API key. Use a FieldReference to inject from a secret |
    | `max_tokens`     | `integer`       | No       | `8192`  | Maximum tokens in responses                                     |
    | `temperature`    | `float`         | No       | —       | Sampling temperature (0.0–1.0)                                  |
    | `top_p`          | `float`         | No       | —       | Nucleus sampling parameter                                      |
    | `top_k`          | `integer`       | No       | —       | Top-k sampling parameter                                        |
    | `stop_sequences` | `list[string]`  | No       | —       | Sequences that stop generation                                  |

    ### Outputs

    | Field  | Type     | Description                                      |
    | ------ | -------- | ------------------------------------------------ |
    | `spec` | `object` | Serialized model spec for runtime reconstruction |

    ### Example

    ```yaml theme={"theme":{"light":"min-light","dark":"min-dark"}}
    provider: agno
    resource: models/anthropic
    name: claude
    config:
      id: claude-sonnet-4-20250514
      api_key:
        provider: pragma
        resource: secret
        name: anthropic-key
        field: outputs.ANTHROPIC_API_KEY
      max_tokens: 4096
      temperature: 0.7
    ```
  </Tab>

  <Tab title="OpenAI (GPT)">
    ## `agno/models/openai`

    Configures an OpenAI model.

    ### Config

    | Field               | Type                       | Required | Default | Description                                                  |
    | ------------------- | -------------------------- | -------- | ------- | ------------------------------------------------------------ |
    | `id`                | `string`                   | **Yes**  | —       | Model identifier (e.g., `gpt-4o`)                            |
    | `api_key`           | `Field[string]`            | **Yes**  | —       | OpenAI API key. Use a FieldReference to inject from a secret |
    | `max_tokens`        | `integer`                  | No       | —       | Maximum tokens in responses                                  |
    | `temperature`       | `float`                    | No       | —       | Sampling temperature (0.0–2.0)                               |
    | `top_p`             | `float`                    | No       | —       | Nucleus sampling parameter                                   |
    | `frequency_penalty` | `float`                    | No       | —       | Frequency penalty (−2.0 to 2.0)                              |
    | `presence_penalty`  | `float`                    | No       | —       | Presence penalty (−2.0 to 2.0)                               |
    | `seed`              | `integer`                  | No       | —       | Random seed for deterministic outputs                        |
    | `stop`              | `string` or `list[string]` | No       | —       | Stop sequences                                               |
    | `timeout`           | `float`                    | No       | —       | Request timeout in seconds                                   |
    | `max_retries`       | `integer`                  | No       | —       | Maximum retry attempts                                       |
    | `organization`      | `string`                   | No       | —       | OpenAI organization ID                                       |
    | `base_url`          | `string`                   | No       | —       | Custom base URL (for OpenAI-compatible APIs)                 |

    ### Outputs

    | Field  | Type     | Description                                      |
    | ------ | -------- | ------------------------------------------------ |
    | `spec` | `object` | Serialized model spec for runtime reconstruction |

    ### Example

    ```yaml theme={"theme":{"light":"min-light","dark":"min-dark"}}
    provider: agno
    resource: models/openai
    name: gpt4
    config:
      id: gpt-4o
      api_key:
        provider: pragma
        resource: secret
        name: openai-key
        field: outputs.OPENAI_API_KEY
      temperature: 0.5
    ```
  </Tab>
</Tabs>

## Dependencies

**Depends on:** Nothing directly, but `api_key` typically uses a FieldReference to a `pragma/secret` resource.

**Depended on by:**

* `agno/agent` — as the required model
* `agno/team` — as the optional team lead model
* `agno/memory/manager` — as the optional classification model

## Notes

* Both model types are **stateless** — they wrap configuration without making API calls during creation.
* The `api_key` field supports FieldReferences for injecting secrets from a `pragma/secret` resource.
* The `base_url` field on the OpenAI model allows you to use OpenAI-compatible APIs (e.g., Azure OpenAI, local models).