class LittleGhost::Providers::LMStudio

LMStudio connects LittleGhost to an LM Studio server through its OpenAI-compatible API. Generation, embeddings, retries, cancellation, and deadlines use the shared compatible transport.

provider = LittleGhost::Providers::LMStudio.new(
  model: "google/gemma-3-4b",
  allow_insecure_http: true
)

The default endpoint is LM Studio’s local server. Pass api_key when the server requires authentication. Plain HTTP remains an explicit opt-in.

Constants

DEFAULT_BASE_URL: Default endpoint for LM Studio’s OpenAI-compatible API.

Public Class Methods

new (api_key: DEFAULT_API_KEY, base_url: DEFAULT_BASE_URL, **arguments)

Source

# File lib/little_ghost/providers/lm_studio.rb, line 27
def initialize(api_key: DEFAULT_API_KEY, base_url: DEFAULT_BASE_URL, **arguments)
  super
end

Configures an LM Studio client. All OpenAICompatible options apply. The placeholder API key is suitable only when LM Studio authentication is disabled.

Calls superclass method

Public Instance Methods

capabilities (metadata: {})

Source

# File lib/little_ghost/providers/lm_studio.rb, line 34
def capabilities(metadata: {})
  parameters = metadata[:supported_parameters] || metadata["supported_parameters"]
  return super unless parameters.is_a?(Array)

  values = parameters.map(&:to_s)
  ModelCapabilities.new(
    native_structured_output: values.include?("structured_outputs"),
    tools: values.include?("tools"),
    tool_choice: values.include?("tool_choice"),
    supported_parameters: values
  )
end

Uses refreshed LM Studio model metadata when available. Before a catalog refresh, the compatible endpoint retains its permissive capability contract.

Calls superclass method