llm.rb is a zero-dependency Ruby toolkit for Large Language Models that includes OpenAI, Gemini, Anthropic, Ollama, and LlamaCpp. It’s fast, simple and composable – with full support for chat, tool calling, audio, images, files, and JSON Schema generation.
- ✅ A single unified interface for multiple providers
- 📦 Zero dependencies outside Ruby's standard library
- 🚀 Optimized for performance and low memory usage
- 🔌 Retrieve models dynamically for introspection and selection
- 🧠 Stateless and stateful chat via completions and responses API
- 🤖 Tool calling and function execution
- 🗂️ JSON Schema support for structured, validated responses
- 🗣️ Text-to-speech, transcription, and translation
- 🖼️ Image generation, editing, and variation support
- 📎 File uploads and prompt-aware file interaction
- 💡 Multimodal prompts (text, images, PDFs, URLs, files)
- 🧮 Text embeddings and vector support
1. Tools: "system" function
2. Files: import at boot time
3. Files: import at runtime
All providers inherit from LLM::Provider – they share a common interface and set of functionality. Each provider can be instantiated using an API key (if required) and an optional set of configuration options via the singleton methods of LLM. For example:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: "yourapikey")
llm = LLM.gemini(key: "yourapikey")
llm = LLM.anthropic(key: "yourapikey")
llm = LLM.ollama(key: nil)
llm = LLM.llamacpp(key: nil)
llm = LLM.voyageai(key: "yourapikey")This example uses the stateless chat completions API that all providers support. A similar example for OpenAI's stateful responses API is available in the docs/ directory.
The following example enables lazy mode for a LLM::Chat object by entering into a conversation where messages are buffered and sent to the provider only when necessary. Both lazy and non-lazy conversations maintain a message thread that can be reused as context throughout a conversation. The example captures the spirit of llm.rb by demonstrating how objects cooperate together through composition, and it uses the stateless chat completions API that all LLM providers support:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
bot = LLM::Chat.new(llm).lazy
msgs = bot.chat do |prompt|
prompt.system File.read("./share/llm/prompts/system.txt")
prompt.user "Tell me the answer to 5 + 15"
prompt.user "Tell me the answer to (5 + 15) * 2"
prompt.user "Tell me the answer to ((5 + 15) * 2) / 10"
end
# At this point, we execute a single request
msgs.each { print "[#{_1.role}] ", _1.content, "\n" }
##
# [system] You are my math assistant.
# I will provide you with (simple) equations.
# You will provide answers in the format "The answer to <equation> is <answer>".
# I will provide you a set of messages. Reply to all of them.
# A message is considered unanswered if there is no corresponding assistant response.
#
# [user] Tell me the answer to 5 + 15
# [user] Tell me the answer to (5 + 15) * 2
# [user] Tell me the answer to ((5 + 15) * 2) / 10
#
# [assistant] The answer to 5 + 15 is 20.
# The answer to (5 + 15) * 2 is 40.
# The answer to ((5 + 15) * 2) / 10 is 4.All LLM providers except Anthropic allow a client to describe the structure of a response that a LLM emits according to a schema that is described by JSON. The schema lets a client describe what JSON object (or value) an LLM should emit, and the LLM will abide by the schema. See also: JSON Schema website. We will use the llmrb/json-schema library for the sake of the examples – the interface is designed so you could drop in any other library in its place:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
schema = llm.schema.object({fruit: llm.schema.string.enum("Apple", "Orange", "Pineapple")})
bot = LLM::Chat.new(llm, schema:).lazy
bot.chat "Your favorite fruit is Pineapple", role: :system
bot.chat "What fruit is your favorite?", role: :user
bot.messages.find(&:assistant?).content! # => {fruit: "Pineapple"}
schema = llm.schema.object({answer: llm.schema.integer.required})
bot = LLM::Chat.new(llm, schema:).lazy
bot.chat "Tell me the answer to ((5 + 5) / 2)", role: :user
bot.messages.find(&:assistant?).content! # => {answer: 5}
schema = llm.schema.object({probability: llm.schema.number.required})
bot = LLM::Chat.new(llm, schema:).lazy
bot.chat "Does the earth orbit the sun?", role: :user
bot.messages.find(&:assistant?).content! # => {probability: 1}The OpenAI, Anthropic, Gemini and Ollama providers support a powerful feature known as tool calling, and although it is a little complex to understand at first, it can be powerful for building agents. The following example demonstrates how we can define a local function (which happens to be a tool), and OpenAI can then detect when we should call the function.
The LLM::Chat#functions method returns an array of functions that can be called after sending a message and it will only be populated if the LLM detects a function should be called. Each function corresponds to an element in the "tools" array. The array is emptied after a function call, and potentially repopulated on the next message.
The following example defines an agent that can run system commands based on natural language, and it is only intended to be a fun demo of tool calling - it is not recommended to run arbitrary commands from a LLM without sanitizing the input first :) Without further ado:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
tool = LLM.function(:system) do |fn|
fn.description "Run a shell command"
fn.params do |schema|
schema.object(command: schema.string.required)
end
fn.define do |params|
ro, wo = IO.pipe
re, we = IO.pipe
Process.wait Process.spawn(params.command, out: wo, err: we)
[wo,we].each(&:close)
{stderr: re.read, stdout: ro.read}
end
end
bot = LLM::Chat.new(llm, tools: [tool]).lazy
bot.chat "Your task is to run shell commands via a tool.", role: :system
bot.chat "What is the current date?", role: :user
bot.chat bot.functions.map(&:call) # report return value to the LLM
bot.chat "What operating system am I running? (short version please!)", role: :user
bot.chat bot.functions.map(&:call) # report return value to the LLM
##
# {stderr: "", stdout: "Thu May 1 10:01:02 UTC 2025"}
# {stderr: "", stdout: "FreeBSD"}Some but not all providers implement audio generation capabilities that
can create speech from text, transcribe audio to text, or translate
audio to text (usually English). The following example uses the OpenAI provider
to create an audio file from a text prompt. The audio is then moved to
${HOME}/hello.mp3 as the final step:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
res = llm.audio.create_speech(input: "Hello world")
IO.copy_stream res.audio, File.join(Dir.home, "hello.mp3")The following example transcribes an audio file to text. The audio file
(${HOME}/hello.mp3) was theoretically created in the previous example,
and the result is printed to the console. The example uses the OpenAI
provider to transcribe the audio file:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
res = llm.audio.create_transcription(
file: File.join(Dir.home, "hello.mp3")
)
print res.text, "\n" # => "Hello world."The following example translates an audio file to text. In this example
the audio file (${HOME}/bomdia.mp3) is theoretically in Portuguese,
and it is translated to English. The example uses the OpenAI provider,
and at the time of writing, it can only translate to English:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
res = llm.audio.create_translation(
file: File.join(Dir.home, "bomdia.mp3")
)
print res.text, "\n" # => "Good morning."Some but not all LLM providers implement image generation capabilities that
can create new images from a prompt, or edit an existing image with a
prompt. The following example uses the OpenAI provider to create an
image of a dog on a rocket to the moon. The image is then moved to
${HOME}/dogonrocket.png as the final step:
#!/usr/bin/env ruby
require "llm"
require "open-uri"
require "fileutils"
llm = LLM.openai(key: ENV["KEY"])
res = llm.images.create(prompt: "a dog on a rocket to the moon")
res.urls.each do |url|
FileUtils.mv OpenURI.open_uri(url).path,
File.join(Dir.home, "dogonrocket.png")
endThe following example is focused on editing a local image with the aid
of a prompt. The image (/images/cat.png) is returned to us with the cat
now wearing a hat. The image is then moved to ${HOME}/catwithhat.png as
the final step:
#!/usr/bin/env ruby
require "llm"
require "open-uri"
require "fileutils"
llm = LLM.openai(key: ENV["KEY"])
res = llm.images.edit(
image: "/images/cat.png",
prompt: "a cat with a hat",
)
res.urls.each do |url|
FileUtils.mv OpenURI.open_uri(url).path,
File.join(Dir.home, "catwithhat.png")
endThe following example is focused on creating variations of a local image.
The image (/images/cat.png) is returned to us with five different variations.
The images are then moved to ${HOME}/catvariation0.png, ${HOME}/catvariation1.png
and so on as the final step:
#!/usr/bin/env ruby
require "llm"
require "open-uri"
require "fileutils"
llm = LLM.openai(key: ENV["KEY"])
res = llm.images.create_variation(
image: "/images/cat.png",
n: 5
)
res.urls.each.with_index do |url, index|
FileUtils.mv OpenURI.open_uri(url).path,
File.join(Dir.home, "catvariation#{index}.png")
endMost LLM providers provide a Files API where you can upload files that can be referenced from a prompt and llm.rb has first-class support for this feature. The following example uses the OpenAI provider to describe the contents of a PDF file after it has been uploaded. The file (an instance of LLM::Response::File) is passed directly to the chat method, and generally any object a prompt supports can be given to the chat method:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
bot = LLM::Chat.new(llm).lazy
file = llm.files.create(file: "/documents/openbsd_is_awesome.pdf")
bot.chat(file)
bot.chat("What is this file about?")
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }
##
# [assistant] This file is about OpenBSD, a free and open-source Unix-like operating system
# based on the Berkeley Software Distribution (BSD). It is known for its
# emphasis on security, code correctness, and code simplicity. The file
# contains information about the features, installation, and usage of OpenBSD.Generally all providers accept text prompts but some providers can
also understand URLs, and various file types (eg images, audio, video,
etc). The llm.rb approach to multimodal prompts is to let you pass URI
objects to describe links, LLM::File | LLM::Response::File objects
to describe files, String objects to describe text blobs, or an array
of the aforementioned objects to describe multiple objects in a single
prompt. Each object is a first class citizen that can be passed directly
to a prompt:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
bot = LLM::Chat.new(llm).lazy
bot.chat [URI("https://example.com/path/to/image.png"), "Describe the image in the link"]
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }
file = llm.files.create(file: "/documents/openbsd_is_awesome.pdf")
bot.chat [file, "What is this file about?"]
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }
bot.chat [LLM.File("/images/puffy.png"), "What is this image about?"]
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }
bot.chat [LLM.File("/images/beastie.png"), "What is this image about?"]
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }The
LLM::Provider#embed
method generates a vector representation of one or more chunks
of text. Embeddings capture the semantic meaning of text –
a common use-case for them is to store chunks of text in a
vector database, and then to query the database for semantically
similar text. These chunks of similar text can then support the
generation of a prompt that is used to query a large language model,
which will go on to generate a response:
#!/usr/bin/env ruby
require "llm"
llm = LLM.openai(key: ENV["KEY"])
res = llm.embed(["programming is fun", "ruby is a programming language", "sushi is art"])
print res.class, "\n"
print res.embeddings.size, "\n"
print res.embeddings[0].size, "\n"
##
# LLM::Response::Embedding
# 3
# 1536Almost all LLM providers provide a models endpoint that allows a client to query the list of models that are available to use. The list is dynamic, maintained by LLM providers, and it is independent of a specific llm.rb release. LLM::Model objects can be used instead of a string that describes a model name (although either works). Let's take a look at an example:
#!/usr/bin/env ruby
require "llm"
##
# List all models
llm = LLM.openai(key: ENV["KEY"])
llm.models.all.each do |model|
print "model: ", model.id, "\n"
end
##
# Select a model
model = llm.models.all.find { |m| m.id == "gpt-3.5-turbo" }
bot = LLM::Chat.new(llm, model:)
bot.chat "Hello #{model.id} :)"
bot.messages.select(&:assistant?).each { print "[#{_1.role}] ", _1.content, "\n" }The README tries to provide a high-level overview of the library. For everything else there's the API reference. It covers classes and methods that the README glances over or doesn't cover at all. The API reference is available at 0x1eef.github.io/x/llm.rb.
The docs/ directory contains some additional documentation that didn't quite make it into the README. It covers the design guidelines that the library follows, some strategies for memory management, and other provider-specific features.
An extensible, developer-oriented command line utility that is powered by llm.rb and serves as a demonstration of the library's capabilities. The demo section has a number of GIF previews might be especially interesting.
llm.rb can be installed via rubygems.org:
gem install llm.rb
