MAX is now source available, with a simplified license
Modular 26.7: MAX is now source available
One open foundation for AI, on your choice of hardware.
AI shouldn’t depend on a single hardware vendor. We’re building MAX so you can build and deploy AI on any hardware platform without vendor lock-in.
As of 26.7, MAX is source available. The graph compiler, runtime and hardware backends join what we’ve already opened: the kernels, models, serving infrastructure, and the Mojo standard library + compiler.
What the new MAX License means for you
You can use MAX, modify it and ship commercial products with it, with no licensing fees or royalties. The few restrictions are there to keep MAX from splitting into incompatible versions, so everyone keeps building on the same foundation.
Building with partners
Our new alliance program gives hardware vendors and partners a shared way to build on MAX together, with direct engineering support, early access to model checkpoints, roadmap visibility, and more.
Also in 26.7:
- Eager mode makes debugging custom ops faster.
- Decision model support adds models like Decider 2B.
- Mojo 1.2 adds pattern matching as Mojo keeps growing into a systems programming language.
Questions about the license? Chris Lattner and Tim Davis are hosting a live Q&A on November 10 at 10 AM PT.RSVP here.
Day zero support for MiniMax M3.1 Flash
MiniMax M3.1 Flash (aka Space Bunny) is live on Modular Cloud, day zero. Among open models, it’s the strongest for cowork automation, with SOTA-level coding and engineering performance. It can ship a fix in a real codebase, then update Jira and Slack afterwards.
We worked with the MiniMax team ahead of launch, so M3.1 Flash is already serving production traffic on Modular Cloud.
The Modular Team 🔥
