How to configure different LLM providers

This guide shows you how to configure and use different LLM providers with LangChainGo.

OpenAI

Basic configuration

import "github.com/tmc/langchaingo/llms/openai"

// Using environment variable OPENAI_API_KEY
llm, err := openai.New()

// Or with explicit API key
llm, err := openai.New(openai.WithToken("your-api-key"))

Advanced configuration

llm, err := openai.New(
    openai.WithToken("your-api-key"),
    openai.WithModel("gpt-4"), // Specify model
    openai.WithBaseURL("https://custom-endpoint.com"), // Custom endpoint
    openai.WithOrganization("org-id"), // Organization ID
    openai.WithAPIVersion("2023-12-01"), // API version
)

Azure OpenAI

import "github.com/tmc/langchaingo/llms/openai"

llm, err := openai.New(
    openai.WithToken("your-azure-api-key"),
    openai.WithBaseURL("https://your-resource.openai.azure.com"),
    openai.WithAPIVersion("2023-12-01-preview"),
    openai.WithAPIType(openai.APITypeAzure),
)

Anthropic

Basic configuration

import "github.com/tmc/langchaingo/llms/anthropic"

// Using environment variable ANTHROPIC_API_KEY
llm, err := anthropic.New()

// Or with explicit API key
llm, err := anthropic.New(anthropic.WithToken("your-api-key"))

Model selection

llm, err := anthropic.New(
    anthropic.WithModel("claude-3-opus-20240229"),
    anthropic.WithToken("your-api-key"),
)

Google AI (Gemini)

Basic configuration

import "github.com/tmc/langchaingo/llms/googleai"

// Using environment variable GOOGLE_API_KEY
llm, err := googleai.New(context.Background())

// Or with explicit API key
llm, err := googleai.New(
    context.Background(),
    googleai.WithAPIKey("your-api-key"),
)

Model configuration

llm, err := googleai.New(
    context.Background(),
    googleai.WithDefaultModel("gemini-pro"),
    googleai.WithAPIKey("your-api-key"),
)

Vertex AI

Basic configuration

import "github.com/tmc/langchaingo/llms/vertexai"

llm, err := vertexai.New(
    context.Background(),
    vertexai.WithProjectID("your-project-id"),
    vertexai.WithLocation("us-central1"),
)

With service account

llm, err := vertexai.New(
    context.Background(),
    vertexai.WithProjectID("your-project-id"),
    vertexai.WithLocation("us-central1"),
    vertexai.WithCredentialsFile("path/to/service-account.json"),
)

Local Models (Ollama)

Basic configuration

import "github.com/tmc/langchaingo/llms/ollama"

// Default configuration (localhost:11434)
llm, err := ollama.New(ollama.WithModel("llama2"))

// Custom server
llm, err := ollama.New(
    ollama.WithServerURL("http://custom-server:11434"),
    ollama.WithModel("codellama"),
)

Hugging Face

Basic configuration

import "github.com/tmc/langchaingo/llms/huggingface"

// Using environment variable HF_TOKEN
llm, err := huggingface.New()

// Or with explicit token
llm, err := huggingface.New(huggingface.WithToken("your-hf-token"))

Model selection

llm, err := huggingface.New(
    huggingface.WithModel("microsoft/DialoGPT-medium"),
    huggingface.WithToken("your-hf-token"),
)

Environment variables

Set up your environment with the appropriate API keys:

# OpenAI
export OPENAI_API_KEY="sk-..."

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."

# Google AI
export GOOGLE_API_KEY="AI..."

# Hugging Face
export HF_TOKEN="hf_..."

# Vertex AI (using Application Default Credentials)
export GOOGLE_APPLICATION_CREDENTIALS="path/to/service-account.json"

Provider-specific features

OpenAI functions

tools := []openai.Tool{
    {
        Type: "function",
        Function: openai.FunctionDefinition{
            Name:        "get_weather",
            Description: "Get current weather",
            Parameters: map[string]any{
                "type": "object",
                "properties": map[string]any{
                    "location": map[string]any{
                        "type":        "string",
                        "description": "City name",
                    },
                },
                "required": []string{"location"},
            },
        },
    },
}

response, err := llm.GenerateContent(ctx, messages, llms.WithTools(tools))

Anthropic system messages

messages := []llms.MessageContent{
    llms.TextParts(llms.ChatMessageTypeSystem, "You are a helpful assistant."),
    llms.TextParts(llms.ChatMessageTypeHuman, "Hello!"),
}

Streaming responses

// Works with most providers
response, err := llm.GenerateContent(
    ctx, 
    messages, 
    llms.WithStreamingFunc(func(ctx context.Context, chunk []byte) error {
        fmt.Print(string(chunk))
        return nil
    }),
)

Error handling

response, err := llm.GenerateContent(ctx, messages)
if err != nil {
    // Check for specific error types
    if errors.Is(err, llms.ErrRateLimit) {
        // Handle rate limiting
        time.Sleep(time.Second * 60)
        // Retry...
    } else if errors.Is(err, llms.ErrQuotaExceeded) {
        // Handle quota exceeded
        log.Fatal("API quota exceeded")
    } else {
        // Handle other errors
        log.Printf("LLM error: %v", err)
    }
}

Best practices

1. Use environment variables: Store API keys securely in environment variables
2. Handle rate limits: Implement retry logic with exponential backoff
3. Model selection: Choose the right model for your use case and budget
4. Error handling: Implement robust error handling for different failure modes
5. Resource management: Use context for timeouts and cancellation
6. Testing: Use mock providers for testing (see testing guide)

Provider comparison

Provider	Strengths	Use cases
OpenAI	High quality, function calling	General purpose, agents
Anthropic	Safety, long context	Research, content analysis
Google AI	Free tier, fast	Experimentation, mobile apps
Vertex AI	Enterprise features	Production, compliance
Ollama	Privacy, offline	Local development, sensitive data
Hugging Face	Open models, variety	Research, experimentation