Class: LLM::Gemini
- Includes:
- RequestAdapter
- Defined in:
- lib/llm/providers/gemini.rb,
lib/llm/providers/gemini/audio.rb,
lib/llm/providers/gemini/files.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,
lib/llm/providers/gemini/request_adapter.rb,
lib/llm/providers/gemini/response_adapter.rb
Overview
The Gemini class implements a provider for [Gemini](ai.google.dev/). 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.
Defined Under Namespace
Modules: RequestAdapter, ResponseAdapter Classes: Audio, ErrorHandler, Files, Images, Models, StreamParser
Constant Summary collapse
- HOST =
"generativelanguage.googleapis.com"
Instance Method Summary collapse
-
#assistant_role ⇒ String
Returns the role of the assistant in the conversation.
-
#audio ⇒ LLM::Gemini::Audio
Provides an interface to Gemini’s audio API.
-
#complete(prompt, params = {}) ⇒ LLM::Response
Provides an interface to the chat completions API.
-
#default_model ⇒ String
Returns the default model for chat completions.
-
#embed(input, model: "text-embedding-004", **params) ⇒ LLM::Response
Provides an embedding.
-
#files ⇒ LLM::Gemini::Files
Provides an interface to Gemini’s file management API.
-
#images ⇒ see LLM::Gemini::Images
Provides an interface to Gemini’s image generation API.
-
#initialize ⇒ Gemini
constructor
A new instance of Gemini.
-
#models ⇒ LLM::Gemini::Models
Provides an interface to Gemini’s models API.
- #server_tools ⇒ String => LLM::ServerTool
-
#web_search(query:) ⇒ LLM::Response
A convenience method for performing a web search using the Google Search tool.
Methods included from RequestAdapter
Methods inherited from Provider
#chat, clients, #inspect, #moderations, #respond, #responses, #schema, #server_tool, #vector_stores, #with
Constructor Details
Instance Method Details
#assistant_role ⇒ 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 |
#audio ⇒ LLM::Gemini::Audio
Provides an interface to Gemini’s audio API
87 88 89 |
# File 'lib/llm/providers/gemini.rb', line 87 def audio LLM::Gemini::Audio.new(self) end |
#complete(prompt, params = {}) ⇒ LLM::Response
Provides an interface to the chat completions API
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 |
# File 'lib/llm/providers/gemini.rb', line 66 def complete(prompt, params = {}) params = {role: :user, model: default_model}.merge!(params) tools = resolve_tools(params.delete(:tools)) params = [params, adapt_schema(params), adapt_tools(tools)].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) = [*(params.delete(:messages) || []), LLM::Message.new(role, prompt)] body = LLM.json.dump({contents: adapt()}.merge!(params)) set_body_stream(req, StringIO.new(body)) res = execute(request: req, stream:) ResponseAdapter.adapt(res, type: :completion) .extend(Module.new { define_method(:__tools__) { tools } }) end |
#default_model ⇒ String
Returns the default model for chat completions
125 126 127 |
# File 'lib/llm/providers/gemini.rb', line 125 def default_model "gemini-2.5-flash" end |
#embed(input, model: "text-embedding-004", **params) ⇒ LLM::Response
Provides an embedding
47 48 49 50 51 52 53 54 |
# File 'lib/llm/providers/gemini.rb', line 47 def (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 = LLM.json.dump({content: {parts: [{text: input}]}}) res = execute(request: req) ResponseAdapter.adapt(res, type: :embedding) end |
#files ⇒ LLM::Gemini::Files
Provides an interface to Gemini’s file management API
103 104 105 |
# File 'lib/llm/providers/gemini.rb', line 103 def files LLM::Gemini::Files.new(self) end |
#images ⇒ see LLM::Gemini::Images
Provides an interface to Gemini’s image generation API
95 96 97 |
# File 'lib/llm/providers/gemini.rb', line 95 def images LLM::Gemini::Images.new(self) end |
#models ⇒ LLM::Gemini::Models
Provides an interface to Gemini’s models API
111 112 113 |
# File 'lib/llm/providers/gemini.rb', line 111 def models LLM::Gemini::Models.new(self) end |
#server_tools ⇒ String => LLM::ServerTool
This method includes certain tools that require configuration through a set of options that are easier to set through the LLM::Provider#server_tool method.
136 137 138 139 140 141 142 |
# File 'lib/llm/providers/gemini.rb', line 136 def server_tools { google_search: server_tool(:google_search), code_execution: server_tool(:code_execution), url_context: server_tool(:url_context) } end |
#web_search(query:) ⇒ LLM::Response
A convenience method for performing a web search using the Google Search tool.
149 150 151 |
# File 'lib/llm/providers/gemini.rb', line 149 def web_search(query:) ResponseAdapter.adapt(complete(query, tools: [server_tools[:google_search]]), type: :web_search) end |