Class: LLM::Gemini

Inherits:
Provider show all
Defined in:
lib/llm/providers/gemini.rb,
lib/llm/providers/gemini/audio.rb,
lib/llm/providers/gemini/files.rb,
lib/llm/providers/gemini/format.rb,
lib/llm/providers/gemini/images.rb,
lib/llm/providers/gemini/models.rb,
lib/llm/providers/gemini/error_handler.rb,
lib/llm/providers/gemini/stream_parser.rb

Overview

The Gemini class implements a provider for Gemini. The Gemini provider can accept multiple inputs (text, images, audio, and video). The inputs can be provided inline via the prompt for files under 20MB or via the Gemini Files API for files that are over 20MB.

Examples:

#!/usr/bin/env ruby
require "llm"

llm = LLM.gemini(key: ENV["KEY"])
bot = LLM::Bot.new(llm)
bot.chat ["Tell me about this photo", File.open("/images/horse.jpg", "rb")]
bot.messages.select(&:assistant?).each { print "[#{_1.role}]", _1.content, "\n" }

Defined Under Namespace

Modules: Response Classes: Audio, Files, Images, Models

Constant Summary collapse

HOST = 
"generativelanguage.googleapis.com"

Instance Method Summary collapse

Methods inherited from Provider

#chat, #chat!, clients, #inspect, #moderations, mutex, #respond, #respond!, #responses, #schema, #tool, #vector_stores, #with

Constructor Details

#initialize ⇒ Gemini

Returns a new instance of Gemini.

Parameters:

  • key (String, nil) —

    The secret key for authentication



38
39
40
 
# File 'lib/llm/providers/gemini.rb', line 38

def initialize(**)
  super(host: HOST, **)
end

Instance Method Details

#web_search(query:) ⇒ LLM::Response

A convenience method for performing a web search using the Google Search tool.

Parameters:

  • query (String) —

    The search query.

Returns:



149
150
151
152
 
# File 'lib/llm/providers/gemini.rb', line 149

def web_search(query:)
  complete(query, tools: [tools[:google_search]])
    .extend(LLM::Gemini::Response::WebSearch)
end

#embed(input, model: "text-embedding-004", **params) ⇒ LLM::Response

Provides an embedding

Parameters:

  • input (String, Array<String>) —

    The input to embed

  • model (String) (defaults to: "text-embedding-004") —

    The embedding model to use

  • params (Hash) —

    Other embedding parameters

Returns:



49
50
51
52
53
54
55
56
 
# File 'lib/llm/providers/gemini.rb', line 49

def embed(input, model: "text-embedding-004", **params)
  model = model.respond_to?(:id) ? model.id : model
  path = ["/v1beta/models/#{model}", "embedContent?key=#{@key}"].join(":")
  req = Net::HTTP::Post.new(path, headers)
  req.body = JSON.dump({content: {parts: [{text: input}]}})
  res = execute(request: req)
  LLM::Response.new(res).extend(LLM::Gemini::Response::Embedding)
end

#complete(prompt, params = {}) ⇒ LLM::Response

Provides an interface to the chat completions API

Examples:

llm = LLM.openai(key: ENV["KEY"])
messages = [{role: "system", content: "Your task is to answer all of my questions"}]
res = llm.complete("5 + 2 ?", messages:)
print "[#{res.choices[0].role}]", res.choices[0].content, "\n"

Parameters:

  • prompt (String) —

    The input prompt to be completed

  • params (Hash) (defaults to: {}) —

    The parameters to maintain throughout the conversation. Any parameter the provider supports can be included and not only those listed here.

Returns:

Raises:

See Also:



68
69
70
71
72
73
74
75
76
77
78
79
80
81
 
# File 'lib/llm/providers/gemini.rb', line 68

def complete(prompt, params = {})
  params = {role: :user, model: default_model}.merge!(params)
  params = [params, format_schema(params), format_tools(params)].inject({}, &:merge!).compact
  role, model, stream = [:role, :model, :stream].map { params.delete(_1) }
  action = stream ? "streamGenerateContent?key=#{@key}&alt=sse" : "generateContent?key=#{@key}"
  model.respond_to?(:id) ? model.id : model
  path = ["/v1beta/models/#{model}", action].join(":")
  req  = Net::HTTP::Post.new(path, headers)
  messages = [*(params.delete(:messages) || []), LLM::Message.new(role, prompt)]
  body = JSON.dump({contents: format(messages)}.merge!(params))
  set_body_stream(req, StringIO.new(body))
  res = execute(request: req, stream:)
  LLM::Response.new(res).extend(LLM::Gemini::Response::Completion)
end

#audio ⇒ LLM::Gemini::Audio

Provides an interface to Gemini’s audio API

Returns:

See Also:



87
88
89
 
# File 'lib/llm/providers/gemini.rb', line 87

def audio
  LLM::Gemini::Audio.new(self)
end

#images ⇒ see LLM::Gemini::Images

Provides an interface to Gemini’s image generation API

Returns:

See Also:



95
96
97
 
# File 'lib/llm/providers/gemini.rb', line 95

def images
  LLM::Gemini::Images.new(self)
end

#files ⇒ LLM::Gemini::Files

Provides an interface to Gemini’s file management API

Returns:

See Also:



103
104
105
 
# File 'lib/llm/providers/gemini.rb', line 103

def files
  LLM::Gemini::Files.new(self)
end

#models ⇒ LLM::Gemini::Models

Provides an interface to Gemini’s models API

Returns:

See Also:



111
112
113
 
# File 'lib/llm/providers/gemini.rb', line 111

def models
  LLM::Gemini::Models.new(self)
end

#assistant_role ⇒ String

Returns the role of the assistant in the conversation. Usually “assistant” or “model”

Returns:

  • (String) —

    Returns the role of the assistant in the conversation. Usually “assistant” or “model”



117
118
119
 
# File 'lib/llm/providers/gemini.rb', line 117

def assistant_role
  "model"
end

#default_model ⇒ String

Returns the default model for chat completions

Returns:

  • (String)

See Also:



125
126
127
 
# File 'lib/llm/providers/gemini.rb', line 125

def default_model
  "gemini-2.5-flash"
end

#tools ⇒ String => LLM::Tool

Note:

This method includes certain tools that require configuration through a set of options that are easier to set through the LLM::Provider#tool method.

Returns:

See Also:



136
137
138
139
140
141
142
 
# File 'lib/llm/providers/gemini.rb', line 136

def tools
  {
    google_search: tool(:google_search),
    code_execution: tool(:code_execution),
    url_context: tool(:url_context)
  }
end