Models - LLM Providers

Agno-Go supports multiple LLM providers with a unified interface.

Supported Models

OpenAI

GPT-4o, GPT-4o-mini, GPT-4 Turbo, GPT-3.5 Turbo
Full streaming support
Function calling

Anthropic Claude

Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
Streaming support
Tool use

GLM (智谱AI) ⭐ NEW in v1.0.2

GLM-4, GLM-4V (vision), GLM-3-Turbo
Chinese language optimized
Custom JWT authentication
Function calling support

Ollama

Run models locally (Llama, Mistral, etc.)
Privacy-first
No API costs

OpenAI

Setup

import "github.com/rexleimo/agno-go/pkg/agno/models/openai"

model, err := openai.New("gpt-4o-mini", openai.Config{
    APIKey:      os.Getenv("OPENAI_API_KEY"),
    Temperature: 0.7,
    MaxTokens:   1000,
})

Configuration

type Config struct {
    APIKey      string  // Required: Your OpenAI API key
    BaseURL     string  // Optional: Custom endpoint (default: https://api.openai.com/v1)
    Temperature float64 // Optional: 0.0-2.0 (default: 0.7)
    MaxTokens   int     // Optional: Max response tokens
}

Supported Models

Model	Context	Best For
`gpt-4o`	128K	Most capable, multimodal
`gpt-4o-mini`	128K	Fast, cost-effective
`gpt-4-turbo`	128K	Advanced reasoning
`gpt-3.5-turbo`	16K	Simple tasks, fast

Example

package main

import (
    "context"
    "fmt"
    "log"
    "os"

    "github.com/rexleimo/agno-go/pkg/agno/agent"
    "github.com/rexleimo/agno-go/pkg/agno/models/openai"
)

func main() {
    model, err := openai.New("gpt-4o-mini", openai.Config{
        APIKey:      os.Getenv("OPENAI_API_KEY"),
        Temperature: 0.7,
    })
    if err != nil {
        log.Fatal(err)
    }

    agent, _ := agent.New(agent.Config{
        Name:  "Assistant",
        Model: model,
    })

    output, _ := agent.Run(context.Background(), "Hello!")
    fmt.Println(output.Content)
}

Anthropic Claude

Setup

import "github.com/rexleimo/agno-go/pkg/agno/models/anthropic"

model, err := anthropic.New("claude-3-5-sonnet-20241022", anthropic.Config{
    APIKey:    os.Getenv("ANTHROPIC_API_KEY"),
    MaxTokens: 2048,
})

Configuration

type Config struct {
    APIKey      string  // Required: Your Anthropic API key
    Temperature float64 // Optional: 0.0-1.0
    MaxTokens   int     // Optional: Max response tokens (default: 4096)
}

Supported Models

Model	Context	Best For
`claude-3-5-sonnet-20241022`	200K	Most intelligent, coding
`claude-3-opus-20240229`	200K	Complex tasks
`claude-3-sonnet-20240229`	200K	Balanced performance
`claude-3-haiku-20240307`	200K	Fast responses

Example

package main

import (
    "context"
    "fmt"
    "log"
    "os"

    "github.com/rexleimo/agno-go/pkg/agno/agent"
    "github.com/rexleimo/agno-go/pkg/agno/models/anthropic"
)

func main() {
    model, err := anthropic.New("claude-3-5-sonnet-20241022", anthropic.Config{
        APIKey:    os.Getenv("ANTHROPIC_API_KEY"),
        MaxTokens: 2048,
    })
    if err != nil {
        log.Fatal(err)
    }

    agent, _ := agent.New(agent.Config{
        Name:         "Claude",
        Model:        model,
        Instructions: "You are a helpful assistant.",
    })

    output, _ := agent.Run(context.Background(), "Explain quantum computing")
    fmt.Println(output.Content)
}

GLM (智谱AI)

Setup

import "github.com/rexleimo/agno-go/pkg/agno/models/glm"

model, err := glm.New("glm-4", glm.Config{
    APIKey:      os.Getenv("ZHIPUAI_API_KEY"),  // Format: {key_id}.{key_secret}
    Temperature: 0.7,
    MaxTokens:   1024,
})

Configuration

type Config struct {
    APIKey      string  // Required: API key in format {key_id}.{key_secret}
    BaseURL     string  // Optional: Custom endpoint (default: https://open.bigmodel.cn/api/paas/v4)
    Temperature float64 // Optional: 0.0-1.0
    MaxTokens   int     // Optional: Max response tokens
    TopP        float64 // Optional: Top-p sampling parameter
    DoSample    bool    // Optional: Whether to use sampling
}

Supported Models

Model	Context	Best For
`glm-4`	128K	General conversation, Chinese language
`glm-4v`	128K	Vision tasks, multimodal
`glm-3-turbo`	128K	Fast responses, cost-effective

API Key Format

GLM uses a special API key format with two parts:

{key_id}.{key_secret}

Get your API key at: https://open.bigmodel.cn/

Example

package main

import (
    "context"
    "fmt"
    "log"
    "os"

    "github.com/rexleimo/agno-go/pkg/agno/agent"
    "github.com/rexleimo/agno-go/pkg/agno/models/glm"
    "github.com/rexleimo/agno-go/pkg/agno/tools/calculator"
    "github.com/rexleimo/agno-go/pkg/agno/tools/toolkit"
)

func main() {
    model, err := glm.New("glm-4", glm.Config{
        APIKey:      os.Getenv("ZHIPUAI_API_KEY"),
        Temperature: 0.7,
        MaxTokens:   1024,
    })
    if err != nil {
        log.Fatal(err)
    }

    agent, _ := agent.New(agent.Config{
        Name:         "GLM Assistant",
        Model:        model,
        Toolkits:     []toolkit.Toolkit{calculator.New()},
        Instructions: "You are a helpful AI assistant.",
    })

    // Chinese language support
    output, _ := agent.Run(context.Background(), "你好！请计算 123 * 456")
    fmt.Println(output.Content)
}

Authentication

GLM uses JWT (JSON Web Token) authentication:

API key is parsed into key_id and key_secret
JWT token is generated with HMAC-SHA256 signing
Token is valid for 7 days
Automatically regenerated for each request

This is handled automatically by the SDK.

Ollama (Local Models)

Setup

Install Ollama: https://ollama.ai
Pull a model: ollama pull llama2
Use in Agno-Go:

import "github.com/rexleimo/agno-go/pkg/agno/models/ollama"

model, err := ollama.New("llama2", ollama.Config{
    BaseURL: "http://localhost:11434",  // Ollama server
})

Configuration

type Config struct {
    BaseURL     string  // Optional: Ollama server URL (default: http://localhost:11434)
    Temperature float64 // Optional: 0.0-1.0
}

Supported Models

Any model available in Ollama:

llama2, llama3, llama3.1
mistral, mixtral
codellama, deepseek-coder
qwen2, gemma2

Example

package main

import (
    "context"
    "fmt"
    "log"

    "github.com/rexleimo/agno-go/pkg/agno/agent"
    "github.com/rexleimo/agno-go/pkg/agno/models/ollama"
)

func main() {
    // Make sure Ollama is running and model is pulled
    model, err := ollama.New("llama2", ollama.Config{
        BaseURL: "http://localhost:11434",
    })
    if err != nil {
        log.Fatal(err)
    }

    agent, _ := agent.New(agent.Config{
        Name:  "Local Assistant",
        Model: model,
    })

    output, _ := agent.Run(context.Background(), "What is Go?")
    fmt.Println(output.Content)
}

Model Comparison

Performance

Provider	Speed	Cost	Privacy	Context
OpenAI GPT-4o-mini	⚡⚡⚡	💰	☁️ Cloud	128K
OpenAI GPT-4o	⚡⚡	💰💰💰	☁️ Cloud	128K
Anthropic Claude	⚡⚡	💰💰	☁️ Cloud	200K
GLM-4	⚡⚡⚡	💰	☁️ Cloud	128K
Ollama	⚡	🆓 Free	🏠 Local	Varies

When to Use Each

OpenAI GPT-4o-mini

Development and testing
High-volume applications
Cost-sensitive use cases

OpenAI GPT-4o

Complex reasoning tasks
Multimodal applications
Production systems

Anthropic Claude

Long context needs (200K tokens)
Coding assistance
Complex analysis

GLM-4

Chinese language applications
Domestic deployment requirements
Fast responses with good quality
Cost-effective for Chinese users

Ollama

Privacy requirements
No internet connectivity
Zero API costs
Development/testing

Switching Models

Switching between models is easy:

// OpenAI
openaiModel, _ := openai.New("gpt-4o-mini", openai.Config{
    APIKey: os.Getenv("OPENAI_API_KEY"),
})

// Claude
claudeModel, _ := anthropic.New("claude-3-5-sonnet-20241022", anthropic.Config{
    APIKey: os.Getenv("ANTHROPIC_API_KEY"),
})

// GLM
glmModel, _ := glm.New("glm-4", glm.Config{
    APIKey: os.Getenv("ZHIPUAI_API_KEY"),
})

// Ollama
ollamaModel, _ := ollama.New("llama2", ollama.Config{})

// Use the same agent code
agent, _ := agent.New(agent.Config{
    Model: openaiModel,  // or claudeModel, glmModel, or ollamaModel
})

Advanced Configuration

Temperature

Controls randomness (0.0 = deterministic, 1.0+ = creative):

model, _ := openai.New("gpt-4o-mini", openai.Config{
    Temperature: 0.0,  // Consistent responses
})

model, _ := openai.New("gpt-4o-mini", openai.Config{
    Temperature: 1.5,  // Creative responses
})

Max Tokens

Limit response length:

model, _ := openai.New("gpt-4o-mini", openai.Config{
    MaxTokens: 500,  // Short responses
})

Custom Endpoints

Use compatible APIs:

model, _ := openai.New("gpt-4o-mini", openai.Config{
    BaseURL: "https://your-proxy.com/v1",  // Custom endpoint
    APIKey:  "your-key",
})

Timeout Configuration

Configure request timeout for LLM calls (default: 60 seconds):

// OpenAI
model, _ := openai.New("gpt-4o-mini", openai.Config{
    APIKey:  os.Getenv("OPENAI_API_KEY"),
    Timeout: 30 * time.Second,  // Custom timeout
})

// Anthropic Claude
claude, _ := anthropic.New("claude-3-5-sonnet-20241022", anthropic.Config{
    APIKey:  os.Getenv("ANTHROPIC_API_KEY"),
    Timeout: 45 * time.Second,  // Custom timeout
})

Default Timeout: 60 seconds Minimum: 1 second Maximum: 10 minutes (600 seconds)

Use Cases:

Short timeout (10-20s): Quick queries, fallback scenarios
Medium timeout (30-60s): Standard operations (default)
Long timeout (120-300s): Complex reasoning, large context

Error Handling:

import (
    "context"
    "errors"
    "time"
)

ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()

output, err := agent.Run(ctx, input)
if err != nil {
    if errors.Is(err, context.DeadlineExceeded) {
        // Handle timeout error
        fmt.Println("Request timed out")
    } else {
        // Handle other errors
        fmt.Printf("Error: %v\n", err)
    }
}

Best Practices:

Set timeout based on expected response time
Use context timeout for request-level control
Monitor timeout errors to adjust settings
Consider retry logic for timeout failures

Best Practices

1. Environment Variables

Store API keys securely:

// Good ✅
APIKey: os.Getenv("OPENAI_API_KEY")

// Bad ❌
APIKey: "sk-proj-..." // Hardcoded

2. Error Handling

Always check for errors:

model, err := openai.New("gpt-4o-mini", openai.Config{
    APIKey: os.Getenv("OPENAI_API_KEY"),
})
if err != nil {
    log.Fatalf("Failed to create model: %v", err)
}

3. Model Selection

Choose based on your needs:

// Development: Fast and cheap
devModel, _ := openai.New("gpt-4o-mini", config)

// Production: More capable
prodModel, _ := openai.New("gpt-4o", config)

4. Context Management

Be mindful of context limits:

// For long conversations, clear memory periodically
if messageCount > 50 {
    agent.ClearMemory()
}

Environment Setup

Create a .env file:

bash

# OpenAI
OPENAI_API_KEY=sk-proj-...

# Anthropic
ANTHROPIC_API_KEY=sk-ant-...

# GLM (智谱AI) - Format: {key_id}.{key_secret}
ZHIPUAI_API_KEY=your-key-id.your-key-secret

# Ollama (optional, defaults to localhost)
OLLAMA_BASE_URL=http://localhost:11434

Load in your code:

import "github.com/joho/godotenv"

func init() {
    godotenv.Load()
}

Next Steps

Add Tools to enhance model capabilities
Learn about Memory for conversation history
Build Teams with mixed models
Explore Examples for real-world usage

Simple Agent - OpenAI example
Claude Agent - Anthropic example
GLM Agent - GLM (智谱AI) example
Ollama Agent - Local model example

Models - LLM Providers ​

Supported Models ​

OpenAI ​

Anthropic Claude ​

GLM (智谱AI) ⭐ NEW in v1.0.2 ​

Ollama ​

OpenAI ​

Setup ​

Configuration ​

Supported Models ​

Example ​

Anthropic Claude ​

Setup ​

Configuration ​

Supported Models ​

Example ​

GLM (智谱AI) ​

Setup ​

Configuration ​

Supported Models ​

API Key Format ​

Example ​

Authentication ​

Ollama (Local Models) ​

Setup ​

Configuration ​

Supported Models ​

Example ​

Model Comparison ​

Performance ​

When to Use Each ​

Switching Models ​

Advanced Configuration ​

Temperature ​

Max Tokens ​

Custom Endpoints ​

Timeout Configuration ​

Best Practices ​

1. Environment Variables ​

2. Error Handling ​

3. Model Selection ​

4. Context Management ​

Environment Setup ​

Next Steps ​

Related Examples ​

Models - LLM Providers

Supported Models

OpenAI

Anthropic Claude

GLM (智谱AI) ⭐ NEW in v1.0.2

Ollama

OpenAI

Setup

Configuration

Supported Models

Example

Anthropic Claude

Setup

Configuration

Supported Models

Example

GLM (智谱AI)

Setup

Configuration

Supported Models

API Key Format

Example

Authentication

Ollama (Local Models)

Setup

Configuration

Supported Models

Example

Model Comparison

Performance

When to Use Each

Switching Models

Advanced Configuration

Temperature

Max Tokens

Custom Endpoints

Timeout Configuration

Best Practices

1. Environment Variables

2. Error Handling

3. Model Selection

4. Context Management

Environment Setup

Next Steps

Related Examples