Notice something off about this package? Help us keep the marketplace safe and trustworthy by reporting inappropriate content or behavior.
Report this packageIn this release of Modelplane we've expanded inference beyond the providers supported in previous releases, with Vultr VKE now available as an inference cluster provider and more flexible provider configuration across cluster types.
This release also makes Modelplane easier to integrate with existing AI tooling. The Anthropic Messages API is now supported end to end, including use cases such as Claude Code. Multi-node inference gets improved scheduling support with pod rank injection, while labels and annotations are propagated through composed workloads.
We've also added new deployment recipes for Qwen 2.5 72B, Laguna-S-2.1, and Nemotron-3.5-Lightning on Microsoft AKS and Nebius.
For developers, the Nix-based control plane setup is now complete, with failures surfaced directly from nix run, and local two-cluster end-to-end testing provides a way to validate Modelplane without cloud infrastructure or GPUs.
Check out https://docs.modelplane.ai/getting-started/ to get started with Modelplane.
@ytsarev is the v0.3.0 release MVP! He added local end-to-end test capabilities that let us exercise Modelplane across multiple inference clusters without relying on cloud infrastructure or GPUs. He also added a Laguna-S-2.1 deployment recipe on Nebius and used it to test the new flow end to end.
Thank you very much to @ytsarev for helping make modelplane easier to test and validate! 🙇♂️
Full Changelog: https://github.com/modelplaneai/modelplane/compare/v0.2.0...v0.3.0