What does judgment mean in RubyLLM?

A RubyLLM::Judge subclass declares typed questions and reads back probabilities, choices or scores, so Ruby code can branch on calibrated answers instead of parsing model prose.

Do I need TypeSafe specifically?

No. RubyLLM supports many providers; TypeSafe appears in its provider list for judgment questions, while chat, tools, media and embeddings work across the rest.

How do I install RubyLLM?

Add ruby_llm version 2.1.0 with bundle add ruby_llm, then configure a provider such as config.openai_api_key inside RubyLLM.configure in a script or a Rails initializer.

Does RubyLLM integrate with Rails?

Yes. It ships Rails generators, Active Record persistence, Active Storage attachments, Hotwire streaming and ERB prompt templates in app/prompts, and it also works in plain Ruby.

Can RubyLLM run local models instead of hosted ones?

Yes. Alongside hosted providers like OpenAI, Anthropic, Gemini and Bedrock it supports Ollama, Ollama Cloud, GPUStack and any OpenAI-compatible endpoint, so a judgment can run on a local model.

Is RubyLLM's evaluation tooling released yet?

Evaluations with datasets, model criteria and Ruby assertions are available on the main branch and documented under a development install, not in a released gem.

What license is RubyLLM under?

MIT. It is at version 2.1.0 with 4,449 stars, though some evaluation tooling lives on main rather than in the released gem.

CommunityJev Repos & SDKs

RubyLLM

Ruby-native AI framework whose RubyLLM::Judge class asks typed judgment questions: a probability that a statement holds, choices or scores, answered by the TypeSafe provider alongside 18 other hosted and local providers under one consistent API.

See highlights

Open github.com

Stars: 4.5k
Language: Ruby
License: MIT
Published by: Community
Last verified: Oct 2026

crmne/ruby_llm

User repository on GitHub

View on GitHub

The Ruby-native AI framework. Chats, agents, tools, images, audio, and video through one consistent API, in plain Ruby or Rails.

Forks: 512
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

About this project

Build AI features the Ruby way. RubyLLM is the Ruby-native AI framework. Work with models, tools, and agents through one consistent API, in plain Ruby or Rails. Website · Getting Started · What's New in 2.1. Battle tested at - Fully private work AI. Build with the models you want. Move between hosted and local providers without rewriting your application, or connect an OpenAI-compatible endpoint. Browse models and pricing · Connect a provider.

Quoted from the project's README. Read the full README.

Tagscommunity ruby framework judge

Works forJev for Classification Jev for Scoring

Used inJev for Software & DevTools

Languages: Ruby
Platforms: API
Capabilities: Open sourceRequires API key

Highlights

Example judge: class Urgency < RubyLLM::Judge with a probability question on urgency returns a calibrated probability.
Lists TypeSafe among 19 providers next to OpenAI, Anthropic, Gemini, Bedrock, Ollama and OpenRouter.
4,449 stars; README documents version 2.1.0 with Rails generators and Active Record persistence.
MIT licensed; evaluations with datasets and Ruby assertions are in development on main.

Watch out

MIT licensed. Judge and evaluation features need provider API keys, and some evaluation tooling is on main rather than in a released gem.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

S1 Ruby

Ruby gem that makes System One measurement a language primitive: psi prepares a state, judge/choose/score return calibrated distributions, and ? collapses them for control flow. Ships TypeSafe, protocol (/v1/systemone) and CUA providers behind one interface with its own spec suite.

Jev Repos & SDKs#community#ruby-gem#sdk

Community

innocentdiaz/s1_rubyRuby

tab-jev

A pre-release Python framework for predictions on mixed text and tabular data: Jev Noul and Score questions turn text columns into numeric features, then a tabular foundation model such as TabPFN learns from labeled rows in context. Offers hosted and fully local backends.

Jev Repos & SDKs#community#tabular#predictions

Community

edamame-labs/tab-jevPython

CloJev

Unofficial portable Clojure SDK for TypeSafe System One. It posts state and typed questions to /v1/systemone, lists available models, and returns validated Clojure data with keyword or string answer keys, on a JVM backend or a native Jolt backend.

Jev Repos & SDKs#community#clojure#sdk

Community

antlobach/clojevClojure

jev-sdk-go

Dependency-free Go 1.24+ client for Jev Noul, Choice, and Score questions: ChoiceOf[T] and ScoreOf[L] return the caller's own types and reject labels the question never offered, with validation before sending, retries that honor Retry-After, OpenRouter support, and eleven tested examples.

Jev Repos & SDKs#community#go#sdk

Community

HomayoonAlimohammadi/jev-sdk-goGo

AgentScope

33k

Alibaba's multi-agent platform, which implements native TypeSafe Jev classification through JevClassifierModel for binary, choice and score routing across agent pipelines. Python 3.11+, Apache-2.0, with a FastAPI agent service and web UI.

Jev Integrations & Adapters#community#multi-agent#framework

Community

agentscope-ai/agentscopePython

@receptron/laya

882

A Node and TypeScript package that runs Convai's Laya checkpoint through ONNX Runtime, with the same system_one shape as the Python reference.

Jev Repos & SDKs#community#typescript#open-models

Community

receptron/layaTypeScript

From the Jev guides

Original write-ups on what this entry is for.

The Open Reimplementations: Local Models That Speak System One

Sep 22, 2026

Closed weights have not stopped the ecosystem from rebuilding the shape in the open. API-compatible servers, retrained heads on small models, and inference-time tricks that skip generation entirely.

Jev Repos & SDKs#open-models#research#local

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai