
Luminal
Luminal compiles AI models for GPUs and other chips and runs them as a pay-as-you-go inference API.

Overview
Luminal: an AI inference compiler with a pay-as-you-go cloud
Luminal is an inference company that compiles AI models ahead of time into native code for GPUs and other accelerators, then serves them. It sells the result two ways: Luminal Cloud, a serverless inference API billed per token for text and per image for image models, and a licensed deployment that runs on a company's own cloud or on-premises hardware. The company's compiler and an Inference OS schedule work across mixed CPUs, GPUs and ASICs, and a GitHub repository for the compiler is linked from the site.
Key Features
- Compiler that turns models into hardware-specific code with graph-level optimization
- Inference OS that schedules and load-balances across CPUs, GPUs and ASICs
- Luminal Cloud, a serverless API with batching and scale-to-zero
- Pay-as-you-go pricing with one prepaid balance and one set of API keys
- Licensed cloud or on-premises deployment with engineering support
- Public benchmarks and a cost calculator on the site
Ideal Use Case
An engineering team that serves open models in production, wants lower cost per token without managing GPU fleets, and may later need the same stack inside its own data center.
How Luminal differentiates
Most inference hosts tune a runtime for one family of chips. Luminal treats compilation as the product, so the same models can target Nvidia GPUs and other chips. Its image pricing passes through the model maker's list price with no markup, according to its pricing page. Production credibility: Luminal announced a $25M Series A on 2026-11-12, led by Standard Capital with Y Combinator, Felicis and SagaVC, after a $5.3M seed in 2025.
FAQ
What does Luminal do? Luminal compiles AI models for GPUs and other chips and serves them through a cloud API or licensed software.
How is Luminal priced? The cloud is pay as you go, with text priced per million tokens and images per returned image. There is no subscription or minimum. On-premises licences are quoted individually.
Is Luminal open source? The site links a GitHub repository for the compiler. Check its licence before relying on it.
Who funds Luminal? Standard Capital led a $25M Series A, with Y Combinator, Felicis and SagaVC taking part.
tl;dr
Luminal is a compiler-first inference cloud with pay-as-you-go pricing and an on-prem option, backed by a $25M Series A led by Standard Capital.
Why Use Luminal
FAQ

User Reviews
Similar Tools





