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

# AI Filter Block

> Apply AI-powered filtering to arrays using various models and providers

<Frame>
  <img src="https://mintcdn.com/odella/MNHyQg3-ZFMTYSp3/block-reference/assets/ai_filter_light.png?fit=max&auto=format&n=MNHyQg3-ZFMTYSp3&q=85&s=be0cded5d5825acb5629897b91aae195" alt="AI Filter Block Screenshot" className="block dark:hidden" width="1152" height="722" data-path="block-reference/assets/ai_filter_light.png" />

  <img src="https://mintcdn.com/odella/MNHyQg3-ZFMTYSp3/block-reference/assets/ai_filter_dark.png?fit=max&auto=format&n=MNHyQg3-ZFMTYSp3&q=85&s=8ab81b0501512f64acc0343fc8ddcd4c" alt="AI Filter Block Screenshot" className="hidden dark:block" width="1206" height="798" data-path="block-reference/assets/ai_filter_dark.png" />
</Frame>

## Overview

The AI Filter Block applies a filter operation over an array of values where the filter condition is evaluated by one of the AI models from the 250+ models Odella connects to. This powerful block allows you to leverage AI capabilities to perform complex filtering tasks on your data.

## Inputs

<ParamField path="System Prompt" type="string">
  The system prompt to send to the model. Optional. Used to provide high-level guidance to the AI model.
</ParamField>

<ParamField path="Condition Prompt" type="string" required>
  The condition prompt to filter the array. Required. This prompt describes the criteria that each array element will be evaluated against.
</ParamField>

<ParamField path="Input" type="any[]" required>
  The input array to filter. Required. Can be an array of any data type.
</ParamField>

## Outputs

<ResponseField name="Indicies" type="number[]">
  The index numbers of each element that passed the filter condition.
</ResponseField>

<ResponseField name="Output" type="any[]">
  The filtered input array containing only the elements that passed the condition.
</ResponseField>

## Editor Settings

<ParamField path="AI Model" type="string" default="gpt-3.5-turbo">
  The AI model to use for evaluating filter conditions. Available models are dynamically populated based on the LLM provider configuration.
</ParamField>

<ParamField path="Temperature" type="number" default={0} min={0} max={2}>
  The sampling temperature to use. Values between 0-2. Higher values (0.8) make output more random, lower values (0.2) make it more focused and deterministic.
</ParamField>

<ParamField path="Top P" type="number" default={1} min={0} max={1}>
  Alternative to temperature sampling. Only tokens comprising the top P probability mass are considered. For example, 0.1 means only tokens in the top 10% probability are considered.
</ParamField>

<ParamField path="Use Top P" type="boolean" default={false}>
  Toggle between using top P sampling or temperature sampling.
</ParamField>

<ParamField path="Max Tokens" type="number" default={2048}>
  The maximum number of tokens to generate in each chat completion.
</ParamField>

<ParamField path="Stop" type="string">
  A sequence where the API will stop generating further tokens.
</ParamField>

<ParamField path="Presence Penalty" type="number" min={-2} max={2}>
  Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.
</ParamField>

<ParamField path="Frequency Penalty" type="number" min={-2} max={2}>
  Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.
</ParamField>

<ParamField path="Seed" type="number">
  If specified, the model will attempt to generate deterministic results for repeated requests with the same seed and parameters.
</ParamField>

### Advanced Settings

<ParamField path="Custom Max Tokens" type="number">
  Overrides the max number of tokens a model can support. Leave blank for preconfigured token limits.
</ParamField>

<ParamField path="Cache Responses" type="boolean" default={false}>
  If enabled, requests with the same parameters and messages will be cached for immediate responses without an API call.
</ParamField>

<ParamField path="Use for subgraph partial output" type="boolean" default={false}>
  If enabled, streaming responses from this node will be shown in Subgraph nodes that call this graph.
</ParamField>

## Error Handling

* The block will retry failed attempts up to 3 times with exponential backoff
* Token limits are automatically enforced based on the selected model
* Invalid responses (non TRUE/FALSE) from the model will trigger an error
* The block includes built-in timeout handling and request cancellation support

## Example Usage

1. Connect an array of items to the Input port
2. Add a condition prompt like "Keep only items that are environmentally sustainable products"
3. Optionally add a system prompt for additional context
4. Configure the model and parameters in the settings
5. Connect the Output port to use the filtered results

## Notes

* The block automatically handles token counting and cost tracking
* Responses are strictly evaluated as TRUE/FALSE
* The block supports parallel processing of array elements
* Built-in caching can improve performance for repeated operations