v6.4.8 v0.9.1

medium risk

webiny/webiny-jsAug 11, 2026 → Aug 14, 20263 days between releases

What's New

Adds GGUF support, LoRA kernels, and Muse Glimmer model.

Breaking Changes

  • `logprobs` now returns natural log probabilities instead of log10

New Features

  • True LoRA kernels and dynamic adapter support
  • First-class support for loading GGUFs
  • Support for Muse Glimmer model
  • Support for lmhead/embedding quantization
  • Major improvements to scheduler and concurrent serving performance

Improvements

  • Bumped serde_with dependency to 3.21.0

Bug Fixes

  • Addressed recent correctness regressions in core

Upgrade Notes

The `logprobs` API output format has changed from log10 to natural log, requiring code updates. GGUF model loading is now fully supported.

The Two Releases

v6.4.8

v6.4.8

Aug 11, 2026· 2 months ago

A minor release addressing an access control issue in the file manager application to ensure private files are gated correctly.

Full release notes
v0.9.1

v0.9.1

Aug 14, 2026· 2 months ago

This release introduces significant improvements to the core scheduler, concurrent serving, and support for new models like Muse Glimmer, along with LoRA kernel improvements and GGUF support.

Full release notes