Skip to main content

Configuring Models

Learn how to configure AI models for your agents.

Supported Providers​

Wegent supports multiple AI model providers:

  • OpenAI (GPT-4, GPT-3.5)
  • Anthropic (Claude)
  • Local models (Ollama, LM Studio)
  • Custom providers

Model Configuration​

OpenAI Example​

model:
  provider: openai
  name: gpt-4
  temperature: 0.7
  max_tokens: 4096
  api_key: ${OPENAI_API_KEY}

Anthropic Example​

model:
  provider: anthropic
  name: claude-3-opus
  temperature: 0.5
  max_tokens: 8192
  api_key: ${ANTHROPIC_API_KEY}

Local Model Example​

model:
  provider: ollama
  name: llama2
  endpoint: http://localhost:11434

Configuration Options​

OptionDescriptionDefault
providerModel providerRequired
nameModel nameRequired
temperatureRandomness (0-1)0.7
max_tokensMax response tokens4096
top_pNucleus sampling1.0
api_keyAPI keyFrom env

Environment Variables​

Set API keys via environment:

export OPENAI_API_KEY="your-key"
export ANTHROPIC_API_KEY="your-key"

Best Practices​

  • Use environment variables for API keys
  • Adjust temperature based on task
  • Monitor token usage
  • Test with different models