Not Diamond

AI model routing solution that enhances large language model (LLM) selection for improved accuracy and cost savings.

Visit Site

Visit

AI Models Large Language Models (LLMs)AI Developer Tools Open Source AI Models AI API Prompt Engineering

About Not Diamond

Not Diamond is the world's leading AI model router, offering a comprehensive multi-model framework that accelerates development and boosts accuracy through data-driven model selection. By choosing the optimal model for each query, Not Diamond improves accuracy by up to 25% and cuts costs by as much as tenfold. It enables seamless creation of tailored routing algorithms using any evaluation data and input set, optimizing performance for your specific applications.

How to Use

Install Not Diamond via pip or npm, then deploy it immediately or train a custom router using your existing evaluation datasets to fine-tune model routing for your specific needs.

Features

Supports joint prompt optimization
Built-in privacy by design
Automatic prompt adaptation
Smart, data-driven routing

Use Cases

Selecting the best LLM for tasks like planning, coding, report analysis, and content creation
Enhancing prompt adaptation across multiple LLMs

Best For

DevelopersMachine learning engineersData scientistsAI professionalsResearch teamsEnterprise AI teams

Pros

Automates prompt tuning for diverse models
Reduces operational costs up to 10x
Enhances accuracy by selecting optimal models per query
Integrates easily with existing evaluation workflows
Ensures data privacy with client-side processing and fuzzy hashing

Cons

May introduce latency under 100ms during model calls
Requires evaluation data to train custom routers effectively

Pricing Plans

Choose the perfect plan. All plans include 24/7 support.

Discovery

Free

Includes up to 100,000 API routing requests per month

Get Started

Possibility

$100/month

Plus $0.001 per additional API request after the free 100K

Get Started

Necessity

Custom pricing

Get Started

FAQs

Is Not Diamond a proxy service?

No, Not Diamond is not a proxy. It recommends the best model for each request, which is then called directly via API, gateway, or locally. It is agnostic to your request management system.

How does Not Diamond choose which model to use?

Not Diamond employs a specialized predictive algorithm trained on extensive evaluation data to accurately identify the best-performing LLM for any input.

Can I integrate Not Diamond with my existing data and pipelines?

Yes, it seamlessly works with your current datasets and evaluation pipelines. Upload your data to generate a tailored routing algorithm in minutes.

When should I consider using Not Diamond?

Not Diamond suits all development stages—from initial API integration to large-scale enterprise deployment routing every request for maximum efficiency.

Does Not Diamond support prompt optimization across different models?

Absolutely. It can leverage frameworks like DSPy or SAMMO, or your custom prompts, to learn the best prompt-model combinations for each query.

How does Not Diamond compare to using a single LLM?

It acts as a meta-model ensemble, combining multiple powerful LLMs to outperform individual models in quality while reducing costs and latency.

Will routing with Not Diamond increase my model call latency?

Routing speed is under 100ms. By directing requests to faster models, it can actually improve overall response times. Deploying on your infrastructure further reduces latency.

Is Not Diamond suitable for retrieval-augmented generation (RAG) and agent workflows?

Yes, it excels in RAG and agent scenarios, improving output quality, speed, and reliability as diverse prompts are processed.

What programming languages does Not Diamond support?

It offers SDKs for Python, TypeScript, and a REST API, allowing integration within any tech stack.

Is Not Diamond SOC-2 compliant?

The platform is currently pursuing SOC-2 certification and expects to be fully compliant in 2025.