---
title: "Using Ellmer Chat Models"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Initializing Chat Models in hellmer}
  %\VignetteEncoding{UTF-8}
  %\VignetteEngine{knitr::rmarkdown}
editor_options: 
  markdown: 
    wrap: 72
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

There are two ways to use [ellmer chat
models](https://ellmer.tidyverse.org/reference/index.html) for batch
processing. This flexibility allows you to 1) pass a function that
creates a ready-to-go chat model or 2) pass a chat model object for more
user control.

## Method 1: Passing a Function

The first method is to pass an ellmer chat model function directly. This
is the preferred method because it lets hellmer setup the model,
specifically setting `echo = "none"` for cleaner console output:

``` r
library(hellmer)

# Sequential processing
chat <- chat_sequential(
  chat_model = chat_claude,
  system_prompt = "Reply concisely"
)

# Parallel processing
chat <- chat_future(
  chat_model = chat_claude,
  system_prompt = "Reply concisely",
  workers = 4
)
```

In this case, hellmer will:

1.  Recognize that `chat_model` is a function
2.  Setup the model with `echo = "none"` and any additional parameters
    you provide
3.  Use this newly created model for batch processing

## Method 2: Passing an Object

The second method is to pass a chat model object. This is useful when
you need more control over model configuration or want to reuse an
existing model:

``` r
library(hellmer)

# Create and configure a chat model
claude <- chat_claude(
  model = "claude-3-7-sonnet-latest",
  system_prompt = "Reply concisely",
  echo = "none",
  max_tokens = 1000
)

# Sequential processing
chat <- chat_sequential(chat_model = claude)

# Parallel processing
chat <- chat_future(
  chat_model = claude,
  workers = 4
)
```

In this case, hellmer will:

1.  Recognize that `chat_model` is an object
2.  Use the model as-is with its existing configuration