OpenClaw
Back to Blog

AI Models Selection Guide

2024年1月20日
ClawDeploy Team
AI
Models
Guide

AI Models Selection Guide

ClawDeploy supports multiple AI models through OpenRouter. This guide helps you choose the best model for your needs.

Supported Models

Claude Series (Anthropic)

Claude Sonnet 4

  • Features: Balanced performance and cost
  • Best for: Daily conversations, general tasks
  • Context: 200K tokens
  • Rating: ⭐⭐⭐⭐⭐

Claude Opus 4

  • Features: Most powerful reasoning
  • Best for: Complex tasks, code generation
  • Context: 200K tokens
  • Rating: ⭐⭐⭐⭐

GPT Series (OpenAI)

GPT-4o

  • Features: Multimodal support, fast response
  • Best for: Image understanding, quick conversations
  • Context: 128K tokens
  • Rating: ⭐⭐⭐⭐⭐

GPT-4 Turbo

  • Features: High performance, large context
  • Best for: Long document processing, complex reasoning
  • Context: 128K tokens
  • Rating: ⭐⭐⭐⭐

Open Source Models

Llama 3.1 70B

  • Features: Open source, cost-effective
  • Best for: General conversation, text generation
  • Context: 128K tokens
  • Rating: ⭐⭐⭐⭐

Mixtral 8x22B

  • Features: MoE architecture, efficient inference
  • Best for: Multi-task processing
  • Context: 64K tokens
  • Rating: ⭐⭐⭐

Price Comparison

ModelInput PriceOutput PriceValue
Claude Sonnet 4$3/M$15/M⭐⭐⭐⭐⭐
Claude Opus 4$15/M$75/M⭐⭐⭐
GPT-4o$5/M$15/M⭐⭐⭐⭐
Llama 3.1 70B$0.9/M$0.9/M⭐⭐⭐⭐⭐

Prices are for reference only, check OpenRouter for real-time pricing

Recommendations

Scenario 1: Daily Chat Bot

Recommended: Claude Sonnet 4 or GPT-4o

  • Fast response
  • Strong understanding
  • Moderate cost

Scenario 2: Code Assistant

Recommended: Claude Opus 4 or GPT-4 Turbo

  • High quality code generation
  • Good complex logic understanding

Scenario 3: Cost Sensitive

Recommended: Llama 3.1 70B

  • Open source model
  • Lowest price
  • Sufficient performance for daily use

Scenario 4: Multilingual Support

Recommended: GPT-4o or Claude Sonnet 4

  • Strong multilingual capabilities
  • Deep cultural understanding

Usage Tips

1. System Prompt Optimization

You are a helpful AI assistant. Please answer user questions in a concise and friendly manner.

2. Context Management

  • Regularly clean conversation history
  • Keep key information
  • Avoid duplicate content

3. Error Handling

  • Set reasonable timeout
  • Implement retry mechanism
  • Log errors

Model Switching

In ClawDeploy Dashboard, you can switch models at any time:

  1. Go to deployment details
  2. Click "Settings"
  3. Select a new model
  4. Save and restart

After switching, the Bot will automatically restart with the new model.

Common Issues

Model responding slowly?

  1. Check network connection
  2. Try a faster model (like GPT-4o)
  3. Reduce context length

How to reduce costs?

  1. Use open source models
  2. Optimize prompt length
  3. Implement caching

Model doesn't support certain features?

Different models have different capability limitations. Check the OpenRouter documentation for details.

Summary

Choosing the right AI model depends on your specific needs:

  • Quality first → Claude Opus 4 / GPT-4 Turbo
  • Balanced value → Claude Sonnet 4 / GPT-4o
  • Cost priority → Llama 3.1 70B

For questions, feel free to submit an Issue on GitHub.