For developers at the frontier

Routing achieves SOTA on every benchmark

By recommending the best model for every query, Not Diamond outperforms individual LLMs on accuracy by 5-25% while reducing costs up to 10x.

The most powerful model router ever built

Make the most of every model  with relentless precision and speed.
Train your own  router
You can get started with Not Diamond's base router in less than five minutes. And if you have your own evaluation data, Not Diamond gives you the ability to train your own custom routers optimized to your use case.
Input
Model 1
Model 2
Model 3
Plan a trip itinerary for Niue...
0.98
0.89
0.95
Write a merge sort in python...
0.83
0.95
1.00
Analyze this technical report...
0.93
0.47
0.81
Write a blog post about LDA...
0.56
0.96
0.79
Breathtakingly fast
Select the right model in less time than it takes to stream a single token.
ddddFarthest star in th()s1xn
Farthest star in the universe
Write an essay
Intelligent tradeoffs
Efficiently leverage faster and cheaper models without degrading quality.
Quality Threshold
$0.003
$0.72
Joint prompt optimization support
Program the best prompt for each LLM so you always call the right model with the right prompt. No more manual tweaking and experimentation.
GPT-4o
Summarize this text
Claude 3.5 Sonnet
Distill the essence of this document
Train your own  router
You can use Not Diamond out of the box or train your own router with your existing evaluation data—giving you hyper-personalized model routing optimized to your use case.
Input
Model 1
Model 2
Model 3
Plan a trip itinerary for Niue...
0.98
0.89
0.95
Write a merge sort in python...
0.83
0.95
1.00
Analyze this technical report...
0.93
0.47
0.81
Write a blog post about LDA...
0.56
0.96
0.79
Breathtakingly fast
Select the right model in less time than it takes to stream a single token.
ddddFarthest star in th()s1xn
Farthest star in the universe
Write an essay
Intelligent tradeoffs
Efficiently leverage faster and cheaper models without degrading quality.
Quality Threshold
$0.003
$0.72
Joint prompt optimization support
Program the best prompt for each LLM so you always call the right model with the right prompt. No more manual tweaking and experimentation.
GPT-4o
Summarize this text
Claude 3.5 Sonnet
Distill the essence of this document

Privacy by design

Not Diamond is not a proxy and all requests are made client-side. Enable fuzzy hashing on our API or deploy directly to your infra for maximum security.

Loved by developers

Roberto Dessi
Samaya AI

I believe that cracking model routing is *the* next big challenge with LLMs. If you are developing any service on top of an LLM you probably need Not Diamond and should check it out

Kareem Arab
Principal HQ

Removing the constraint of having to switch between models for different tasks is crucial, and this is how it should be done. Awesome stuff!

Tyler Miracle
South Park Commons

With the choices for model selection ever increasing, Not Diamond has knocked it out of the park with their model router. Give it a try if you find yourself shuffling models in your config in search of the best choice for the problem at hand.

McKay Wrigley
Takeoff AI

In a world with many different models, it makes a *ton* of sense to have a really good LLM router to pick the best model for a given task.

So glad to finally have a steerable model router out there!

Francisco Ingham
Pampa AI

If you use the same model for all your requests, you'll probably need a highly capable model and overpay (cost/latency) for your simpler requests.

Been having fun using Not Diamond for request routing, results are impressive. Awesome tool.

Marc Campbell
Replicated

This is cool.  I'm generally just using chat-gpt on 4o because it's currently loaded and open.  Removing the decision and effort from me and letting a computer do the work to pick the model is right.

Ayush Kalani
Aisera

Not Diamond is the much needed LLM router to automatically route to the best LLM.

With so many LLMs and updates every other day it’s so hard to keep track and update applications. This has been so helpful for cutting through the noise.

Jarod Reyes
Auth0

Incredible work, no longer have to wonder "which model do I use" for this AI project. The team at NotDiamond are among some of the best developers I have ever met. Good work team! 🔥

Takaaki Kakei / 筧剛彰
classmethod

クエリに対する最適なLLMを自動的に決定してくれるAIモデルルーター