Documentation

Welcome to LLM Trust! This guide will help you get up and running in minutes.

What is LLM Trust?

LLM Trust is the developer's platform for discovering, comparing, and reviewing open-source large language models. We provide real benchmarks, community reviews, and side-by-side comparisons so you can find the best model for your project.

Quick Start

1. Browse models — Head to /models to explore 200+ LLMs
2. Create an account — Sign up to save favorites, leave reviews, and submit models
3. Compare — Pick two models and see a detailed side-by-side comparison
4. Contribute — Share your experience by reviewing models you've used

Installation (CLI)

# Install via npm
npm install -g @llmtrust/cli

# Or with Homebrew
brew install llmtrust

# Search for a model
llmt search "code generation"

# Run a model locally
llmt run deepseek-coder-v2

The Models page is your starting point for discovering LLMs.

Search & Filter

• Search bar — Type a model name, architecture, or keyword
• Category badges — Filter by use case: text-generation, code, vision, embedding
• Sort options — Order by popularity, newest, downloads, or name

Model Cards

Each model card shows key information at a glance:

• Parameter count — e.g., "7B", "70B", "8x7B" (MoE)
• Architecture — e.g., llama, mistral, qwen
• Category — Primary use case
• License — Apache 2.0, MIT, Llama Community, etc.
• Download count — Community adoption signal

Model Detail Page

Click any model card to see the full detail page with:

• Complete specifications and benchmarks
• Community reviews and ratings
• One-command CLI run instructions
• HuggingFace download link
• Related models and comparisons

Side-by-side comparison helps you choose between models for your specific needs.

How to Compare

1. Go to /compare
2. Select two models from the dropdowns
3. View the comparison table with specs, benchmarks, and reviews

Direct URL

You can also link directly to a comparison using the URL format:

/compare/[model-a-slug]/vs/[model-b-slug]

# Example
/compare/llama-3-70b-vs-gpt-4/page

What's Compared

• Parameter count and architecture
• Context length and training data
• License and pricing
• Benchmark scores (MMLU, HumanEval, etc.)
• Community ratings and reviews
• Download statistics

Any registered user can submit an open-source model for listing on LLM Trust.

Requirements

• A free LLM Trust account
• The model must be open-source with a valid license
• A public HuggingFace (or equivalent) download URL

Upload Process

1. Go to /models/upload
2. Basic Info — Name, slug (auto-generated), and description
3. Technical — Architecture, parameter count, context length, license, category
4. Links — HuggingFace download URL
5. Details — Format (GGUF, Safetensors, etc.) and tags
6. Preview — Review your model card before submitting

Review Process

All submissions are reviewed by our team before publication to ensure quality and accuracy. You'll receive an email once your model is approved.

LLM Trust provides a REST API for integrating model data into your applications.

Authentication

Generate an API key from your Dashboard → API Keys. Include it in the Authorization header:

Authorization: Bearer llmt_your_api_key_here

Endpoints

Full API documentation with request/response schemas is available at /docs/api.