Skip to content

RobotLab

OpenAI

madbomber/robot_lab

OpenAI Adapter¶

Adapter for GPT models via OpenAI API.

Class: `RobotLab::Adapters::OpenAI`¶

# Automatically used for GPT models
robot = RobotLab.build do
  model "gpt-4o"
end

Supported Models¶

Model	Description
`gpt-4o`	Latest GPT-4 Omni
`gpt-4o-mini`	Fast, efficient GPT-4
`gpt-4-turbo`	GPT-4 Turbo
`o1-preview`	Reasoning model
`o1-mini`	Fast reasoning model

Configuration¶

API Key¶

export OPENAI_API_KEY="sk-..."

Options¶

RobotLab.configure do |config|
  config.adapter_options = {
    openai: {
      base_url: "https://api.openai.com/v1",
      organization: "org-...",
      timeout: 120,
      max_tokens: 4096
    }
  }
end

Azure OpenAI¶

RobotLab.configure do |config|
  config.adapter_options = {
    openai: {
      base_url: "https://your-resource.openai.azure.com",
      api_key: ENV["AZURE_OPENAI_KEY"],
      api_version: "2024-02-15-preview"
    }
  }
end

Features¶

Streaming¶

result = robot.run(state: state) do |event|
  case event
  when :text_delta
    print event.text
  when :tool_call
    puts "Calling: #{event.name}"
  end
end

Tool Use¶

Tools are automatically converted to OpenAI's function calling format:

robot = RobotLab.build do
  model "gpt-4o"

  tool :get_weather do
    description "Get current weather"
    parameter :location, type: :string, required: true
    handler { |location:, **_| WeatherAPI.fetch(location) }
  end
end

JSON Mode¶

robot = RobotLab.build do
  model "gpt-4o"
  template "Always respond with valid JSON."
  # Response format is automatically configured
end

Response Format¶

{
  content: [TextMessage, ...],
  tool_calls: [ToolCallMessage, ...],
  usage: {
    input_tokens: 150,
    output_tokens: 250,
    total_tokens: 400
  },
  stop_reason: "stop"
}

Error Handling¶

begin
  result = robot.run(state: state)
rescue RobotLab::Adapters::RateLimitError => e
  sleep(e.retry_after || 60)
  retry
rescue RobotLab::Adapters::APIError => e
  logger.error("OpenAI API error: #{e.message}")
end

See Also¶