Not Diamond
Per-query LLM router that sends each request to the best model for the price.
About
Not Diamond is an LLM routing startup that came out of stealth in 2024 with a novel router trained on labeled examples of which model performed best on which prompt type. For each incoming request it predicts which model will deliver the best answer at the lowest price, balancing quality, cost, and latency instead of sending everything to one model.
The company raised 2.3 million dollars in initial funding from defy.vc with angels including Jeff Dean, Julien Chaumond, Ion Stoica, and Tom Preston-Werner. In 2026 it repositioned specifically toward routing for coding agents, and its router is used by customers such as OpenRouter and Rootly with reported 20 to 40 percent cost cuts. The service charges per routed token and is closed-source SaaS aimed at enterprise workloads.
What it can do
- Per-query model selection across providers
- Cost, latency, and quality tradeoff controls
- Routing tuned for coding agent workloads
- SOC2 and ISO27001 enterprise posture
- Drop-in OpenAI-compatible routing endpoint
- Usage analytics across routed models
Setup
Sign up for an API key and call the routing endpoint instead of a single model provider.
Prerequisites
A Not Diamond account; existing model provider API keys where needed.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 91; Trending score 72/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.