Luminal Blog

Reducing Image Generation cost with AMD and the Luminal Compiler

With the Luminal compiler and AMD’s MI300x, we show that image generation with Flux.2 Klein 9B can be reduced by 47%

Sep 17 • Reis McMillan

4

May 2026

Hardware Arbitrage with Luminal and Positron AI

Luminal compiles off-the-shelf models to Positron Atlas and Nvidia Blackwell to unlock fast and cheap inference across heterogeneous chips.

May 18 • Jonah Samost

7

1

April 2026

Producing The Perfect Token

The unspoken inference quality gap and how numerics determine if the inference you're paying for is worth it.

Apr 6 • Jonah Samost and Joe Fioti

3

January 2026

Compiling Models to Megakernels

Fine-grained synchronization, deep pipelines, and zero kernel launch overheads, automatically.

Jan 9 • Jonah Samost and Joe Fioti

9

2

November 2025

Announcing our $5.3M Seed Round

Luminal has raised a $5.3M seed round to bring speed-of-light inference to everyone.

Nov 18, 2025 • Jonah Samost

3