You can use LangWatch to trace interactions with local models served by Ollama. Ollama exposes an OpenAI-compatible endpoint, making it easy to integrate with the standard otelopenai middleware.

​
Setup

First, ensure Ollama is running and you have pulled a model. 
# Pull the Llama 3 model
ollama pull llama3
By default, Ollama’s server runs at http://localhost:11434.

​
Example

Configure your openai.Client to point to the local Ollama endpoint. While Ollama doesn’t require an API key, the openai-go library requires one to be set, so you can use a non-empty string like "ollama". Set the gen_ai.system attribute to "ollama" to identify the provider in your LangWatch traces.
The following example assumes you have already configured the LangWatch SDK. See the Go setup guide for details.
package main

import (
	"context"
	"log"

	"github.com/langwatch/langwatch/sdk-go"
	otelopenai "github.com/langwatch/langwatch/sdk-go/instrumentation/openai"
	"github.com/openai/openai-go"
	oaioption "github.com/openai/openai-go/option"
)

func main() {
	ctx := context.Background()

	client := openai.NewClient(
		// Use the default Ollama endpoint
		oaioption.WithBaseURL("http://localhost:11434/v1"),

		// The API key is required but not used by Ollama
		oaioption.WithAPIKey("ollama"),

		// Add the middleware, identifying the system as "ollama"
		oaioption.WithMiddleware(otelopenai.Middleware("my-ollama-app",
			otelopenai.WithGenAISystem("ollama"),
			otelopenai.WithCaptureInput(),
			otelopenai.WithCaptureOutput(),
		)),
	)

	// Make a call to a local model
	response, err := client.Chat.Completions.New(ctx, openai.ChatCompletionNewParams{
		Model: "llama3",
		Messages: []openai.ChatCompletionMessageParamUnion{
			openai.UserMessage("Hello, local model! Write a haiku about Go."),
		},
	})

	if err != nil {
		log.Fatalf("Ollama API call failed: %v", err)
	}

	log.Printf("Response from Ollama: %s", response.Choices[0].Message.Content)
}
This same pattern works for any tool that provides an OpenAI-compatible endpoint for local models, such as LM Studio or vLLM.