GLM-5.3 open weights are now public, and Modular Cloud has Day Zero support.
GLM-5.3 is the most capable open-weights model for coding, with a 50% improvement over GLM-5.2 on @Zai_org's Code Bench.
Try it today on Modular Cloud: console.modular.com/?utm_source=x&…
Heterogeneous compute is here, and it has a single, open software platform.
We open sourced the Mojo compiler and toolchain the same day we hosted #ModCon2026, a defining moment for our community and the broader ecosystem.
"We built Mojo to unlock the next generation of compute
It's official: Ox Alpha is @Zai_org's GLM-5.3-Flash, the first natively multimodal model in the GLM-5 series, with 320B total parameters and 18B active, and a 1M-token context window.
Create a free Modular Cloud account now to get access as soon as our endpoint is live:
Introducing GLM-5.3-Flash
- Leading capabilities at a highly competitive price
- Natively multimodal with a 1M-token context window
- A 320B-A18B model released under the MIT License
- Previously previewed as Ox Alpha, running entirely on Chinese AI chips
Blog:
Two more sessions from #ModCon2026 are live on our YouTube channel today.
Modular Cloud: Performance At Scale is a look at the full stack behind Modular Cloud, our managed inference platform, from kernels to serving to orchestration, now generally available. Watch:
