Chat Models

Fundamental chat model interfaces for invocation, message generation, batching, and configuration.

Langchain.Core.Model Langchain.Provider.Ollama Langchain.Provider.OpenAI

Fundamental chat model interfaces for invocation, message generation, batching, and configuration.

Key Concepts

ChatModel Abstraction: Unified interface for interacting with LLM providers supporting invocation, batch requests, and parameter customization.
Multi-Turn Messages: Structured message sequence (userMessage, assistantMessage, systemMessage) representing conversation history.
Batch Processing: Execute multiple prompt collections concurrently with optimal throughput.

Working Code Example

Compare local execution via Ollama and cloud API execution via OpenAI / OpenRouter. Use the toggle tabs or the global provider switcher in the header to switch:

{-# LANGUAGE OverloadedStrings #-}

module Ollama.Simple (runApp) where

import Control.Monad.Except (runExceptT)
import qualified Data.Map as Map
import qualified Data.Text.IO as T
import qualified Data.Text.Lazy as T
import Langchain.Prelude
import Langchain.PromptTemplate.Prompt

runPromptTemplateExample :: IO ()
runPromptTemplateExample = do
  let items = Map.fromList [("name", "John"), ("age", "25")]
  let inputText = "{name} is of age of {age}, he is only {age}!"
  case renderFStringTemplate items inputText of
    Left err -> T.putStrLn $ errorMessage err
    Right r -> T.putStrLn r

runApp :: IO ()
runApp = do
  runPromptTemplateExample
  let prompt = "Write a poem about functional programming"
  let splittedChars = splitTextRecursive defaultRecursiveCharacterSplitterOps prompt
  mapM_ (T.putStrLn . T.toStrict) splittedChars
  o <- newOllama "gemma3" defaultConfig
  let msg = [(userMessage . T.toStrict) prompt]
  res <- runExceptT $ invoke o msg Nothing
  case res of
    Left err -> T.putStrLn $ errorMessage err
    Right m -> T.putStrLn $ extractMessageText m

  let reqWithOptions =
        withOptions
          (defaultOptions {optTemperature = Just 0.7, optNumCtx = Just 4096})
          (chatRequestFor o msg)
  resWithOptions <- runExceptT $ invoke o msg (Just reqWithOptions)
  case resWithOptions of
    Left err -> T.putStrLn $ errorMessage err
    Right m -> T.putStrLn $ extractMessageText m

  let ques =
        [ "What is Functor in Haskell?"
        , "What is applicative in Haskell?"
        , "What is Monad in Haskell?"
        , "A really long question"
        ]
      msgs = map (\q -> [userMessage q]) ques
  batchRes <- runExceptT $ batch o (take 3 msgs) Nothing
  case batchRes of
    Left err -> T.putStrLn $ errorMessage err
    Right ms -> mapM_ (T.putStrLn . extractMessageText) ms

{-# LANGUAGE OverloadedStrings #-}

module OpenAI.Simple (runApp) where

import Control.Monad.Except (runExceptT)
import Data.Aeson (object, (.=))
import qualified Data.Map as Map
import qualified Data.Text.IO as T
import qualified Data.Text.Lazy as T
import Langchain.Prelude
import Langchain.PromptTemplate.Prompt
import OpenAI.Common (defaultModelName, getOpenRouterModel)

runPromptTemplateExample :: IO ()
runPromptTemplateExample = do
  let items = Map.fromList [("name", "John"), ("age", "25")]
  let inputText = "{name} is of age of {age}, he is only {age}!"
  case renderFStringTemplate items inputText of
    Left err -> T.putStrLn $ errorMessage err
    Right r -> T.putStrLn r

runApp :: IO ()
runApp = do
  runPromptTemplateExample
  let prompt = "Write a poem about functional programming"
  let splittedChars = splitTextRecursive defaultRecursiveCharacterSplitterOps prompt
  mapM_ (T.putStrLn . T.toStrict) splittedChars
  o <- getOpenRouterModel defaultModelName
  let msg = [(userMessage . T.toStrict) prompt]
  res <- runExceptT $ invoke o msg Nothing
  case res of
    Left err -> T.putStrLn $ errorMessage err
    Right m -> T.putStrLn $ extractMessageText m

  let reqWithOptions =
        object
          [ "temperature" .= (0.7 :: Double)
          , "max_tokens" .= (1024 :: Int)
          ]
  resWithOptions <- runExceptT $ invoke o msg (Just reqWithOptions)
  case resWithOptions of
    Left err -> T.putStrLn $ errorMessage err
    Right m -> T.putStrLn $ extractMessageText m

  let ques =
        [ "What is Functor in Haskell?"
        , "What is applicative in Haskell?"
        , "What is Monad in Haskell?"
        , "A really long question"
        ]
      msgs = map (\q -> [userMessage q]) ques
  batchRes <- runExceptT $ batch o (take 3 msgs) Nothing
  case batchRes of
    Left err -> T.putStrLn $ errorMessage err
    Right ms -> mapM_ (T.putStrLn . extractMessageText) ms

Core Types & Functions

ChatModel m => m -> [Message] -> Maybe RequestOptions -> ExceptT LangchainError IO Message

ChatModel m => m -> [[Message]] -> Maybe RequestOptions -> ExceptT LangchainError IO [Message]

Text -> Message

Running This Example

Local Ollama

Ensure your Ollama daemon is running locally with the target model:

ollama run gemma3 # or your desired model
stack run simpleollama

OpenAI / OpenRouter

Ensure your OPENROUTER_API_KEY or OPENAI_API_KEY is exported:

export OPENROUTER_API_KEY="your-api-key"
stack run simpleopenai