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

# SDK Constants

<Tabs items={['TypeScript', 'Python']}>
  <Tab title="TypeScript">
    <Heading level={2} id="ts-llm-attributes">LaminarAttributes</Heading>

    Attribute constants for manual LLM span creation.

    ```typescript theme={null}
    import { LaminarAttributes } from '@lmnr-ai/lmnr';

    span.setAttributes({
      [LaminarAttributes.PROVIDER]: 'anthropic',
      [LaminarAttributes.REQUEST_MODEL]: 'claude-3-5-sonnet',
      [LaminarAttributes.RESPONSE_MODEL]: response.model,
      [LaminarAttributes.INPUT_TOKEN_COUNT]: response.usage.input_tokens,
      [LaminarAttributes.OUTPUT_TOKEN_COUNT]: response.usage.output_tokens,
    });
    ```

    | Constant             | Value                        | Description       |
    | -------------------- | ---------------------------- | ----------------- |
    | `PROVIDER`           | `gen_ai.system`              | LLM provider name |
    | `REQUEST_MODEL`      | `gen_ai.request.model`       | Requested model   |
    | `RESPONSE_MODEL`     | `gen_ai.response.model`      | Actual model used |
    | `INPUT_TOKEN_COUNT`  | `gen_ai.usage.input_tokens`  | Input tokens      |
    | `OUTPUT_TOKEN_COUNT` | `gen_ai.usage.output_tokens` | Output tokens     |
    | `TOTAL_TOKEN_COUNT`  | `llm.usage.total_tokens`     | Total tokens      |
    | `INPUT_COST`         | `gen_ai.usage.input_cost`    | Input cost        |
    | `OUTPUT_COST`        | `gen_ai.usage.output_cost`   | Output cost       |
    | `TOTAL_COST`         | `gen_ai.usage.cost`          | Total cost        |

    ***

    <Heading level={2} id="ts-span-type-trace-type">SpanType / TraceType</Heading>

    ```typescript theme={null}
    import type { SpanType, TraceType } from '@lmnr-ai/lmnr';

    // SpanType:
    // 'DEFAULT' | 'LLM' | 'EXECUTOR' | 'EVALUATOR' | 'HUMAN_EVALUATOR'
    // | 'EVALUATION' | 'TOOL' | 'CACHED'
    //
    // TraceType: 'DEFAULT' | 'EVALUATION'
    ```
  </Tab>

  <Tab title="Python">
    <Heading level={2} id="py-llm-attributes">Attributes</Heading>

    Attribute constants for manual LLM span creation.

    ```python theme={null}
    from lmnr import Laminar, Attributes

    Laminar.set_span_attributes({
        Attributes.PROVIDER: "anthropic",
        Attributes.REQUEST_MODEL: "claude-3-5-sonnet",
        Attributes.RESPONSE_MODEL: response["model"],
        Attributes.INPUT_TOKEN_COUNT: response["usage"]["input_tokens"],
        Attributes.OUTPUT_TOKEN_COUNT: response["usage"]["output_tokens"],
    })
    ```

    | Constant             | Value                        | Description       |
    | -------------------- | ---------------------------- | ----------------- |
    | `PROVIDER`           | `gen_ai.system`              | LLM provider name |
    | `REQUEST_MODEL`      | `gen_ai.request.model`       | Requested model   |
    | `RESPONSE_MODEL`     | `gen_ai.response.model`      | Actual model used |
    | `INPUT_TOKEN_COUNT`  | `gen_ai.usage.input_tokens`  | Input tokens      |
    | `OUTPUT_TOKEN_COUNT` | `gen_ai.usage.output_tokens` | Output tokens     |
    | `TOTAL_TOKEN_COUNT`  | `gen_ai.usage.total_tokens`  | Total tokens      |
    | `INPUT_COST`         | `gen_ai.usage.input_cost`    | Input cost        |
    | `OUTPUT_COST`        | `gen_ai.usage.output_cost`   | Output cost       |
    | `TOTAL_COST`         | `gen_ai.usage.cost`          | Total cost        |
    | `RESPONSE_ID`        | `gen_ai.response.id`         | Response ID       |

    ***

    <Heading level={2} id="py-span-type-trace-type">SpanType / TraceType</Heading>

    ```python theme={null}
    from lmnr import SpanType, TraceType

    # SpanType: DEFAULT, LLM, TOOL, EXECUTOR, EVALUATOR, HUMAN_EVALUATOR, EVALUATION
    #   (PIPELINE is reserved internally)
    # TraceType: DEFAULT, EVALUATION
    ```
  </Tab>
</Tabs>